Psychological treatments for people with epilepsy

Rosa Michaelis; Venus Tang; Sarah J Nevitt; Janelle L Wagner; Avani C Modi; William Curt LaFrance Jr; Laura H Goldstein; Milena Gandy; Rebecca Bresnahan; Kette Valente; Kirsten A Donald; Markus Reuber

doi:10.1002/14651858.CD012081.pub3

Tratamientos psicológicos para los pacientes con epilepsia

Declaraciones de intereses de los autores

Versión publicada: 07 septiembre 2020 Historial de versiones

https://doi.org/10.1002/14651858.CD012081.pub3

Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

Debido al impacto significativo que podría tener la epilepsia sobre la calidad de vida relacionada con la salud (CdVRS) en los individuos con epilepsia y sus familias, hay un creciente interés clínico en los tratamientos psicológicos basados en evidencia, dirigidos a mejorar los desenlaces psicológicos y relacionados con las crisis para este grupo de personas.

Esta es una versión actualizada de la revisión Cochrane original publicada en el número 10, 2017.

Objetivos

Evaluar los efectos de los tratamientos psicológicos para los pacientes con epilepsia en los resultados de la CdVRS.

Métodos de búsqueda

Para esta actualización, se realizaron búsquedas en las siguientes bases de datos el 12 de agosto de 2019 sin restricciones de idioma: La Cochrane Register of Studies (CRS Web),que incluye ensayos controlados aleatorizados o cuasialeatorizados del registro especializado de los Grupos Cochrane de Revisión de Epilepsia, Cochrane Central de Ensayos Controlados (CENTRAL), MEDLINE (Ovid, 1946 a 9 de agosto de 2019), y PsycINFO (EBSCOhost, 1887 en adelante), y de PubMed, EMBASE, ClinicalTrials.gov y la plataforma de registros internacionales de ensayos clínicos (ICTRP) de la Organización Mundial de la Salud. Se examinaron las referencias de los estudios incluidos y las revisiones pertinentes, y se estableció contacto con los investigadores del tema para obtener estudios no publicados.

Criterios de selección

Se consideraron los ensayos controlados aleatorizados (ECA) y los ensayos controlados cuasialeatorizados para esta revisión. La CdVRS fue el desenlace principal. Para la definición operativa de “tratamientos psicológicos”, se incluyó un rango amplio de tratamientos psicológicos basados en habilidades e intervenciones solo educativas diseñadas para mejorar la CdVRS, la frecuencia y la gravedad de las crisis, y las comorbilidades psiquiátricas y conductuales para los adultos y los niños con epilepsia. Estos tratamientos psicológicos se compararon con el tratamiento habitual (TH), un grupo de control activo (como el grupo de apoyo social) o la farmacoterapia antidepresiva.

Obtención y análisis de los datos

Se utilizaron los procedimientos metodológicos estándares previstos por Cochrane.

Resultados principales

Se incluyeron 36 ECA completos con 3526 participantes. De estos, 27 estudios investigaron intervenciones psicológicas basadas en habilidades. Los nueve estudios restantes fueron intervenciones solo educativas. Seis estudios investigaron intervenciones para niños y adolescentes, tres estudios investigaron intervenciones para adolescentes y adultos, y los restantes investigaron intervenciones en adultos. De acuerdo con una homogeneidad clínica y metodológica satisfactoria, se agruparon los datos de 11 estudios (643 participantes) que utilizaron la escala para medir la calidad de vida en la epilepsia Quality of Life in Epilepsy‐31 (QOLIE‐31) u otros listados QOLIE (como el QOLIE‐89 o el QOLIE‐31‐P) convertibles a QOLIE‐31. Se encontraron cambios medios significativos para la puntuación total de la QOLIE‐31 y seis subescalas (bienestar emocional, energía y fatiga, CdV general, preocupación sobre la crisis, efectos de la medicación y función cognitiva). Los cambios medios en la puntuación total de la QOLIE‐31 (mejora media de 5,23 puntos, IC del 95%: 3,02 a 7,44; P < 0,001), y la puntuación global de la CdV (mejora media de 5,95 puntos, IC del 95%: 3,05 a 8,85; P < 0,001) superaron el umbral de cambio mínimo importante (CMI: puntuación total: 4,73 puntos; puntuación de CdV: 5,22 puntos), lo que indica una mejoría posintervención clínicamente significativa en la CdVRS. La certeza de la evidencia aportada por el metanálisis se redujo debido a riesgos graves de sesgo en algunos de los estudios incluidos. En consecuencia, estos resultados aportaron evidencia de certeza moderada de que los tratamientos psicológicos para los adultos con epilepsia pueden mejorar la CdVRS general.

Conclusiones de los autores

Implicaciones para la práctica: Las intervenciones psicológicas basadas en habilidades mejoran la CdVRS en adultos y adolescentes con epilepsia. El uso adyuvante de terapias psicológicas basadas en habilidades para los adultos y adolescentes con epilepsia podría proporcionar beneficios adicionales a la CdVRS cuando se incorporan a un tratamiento centrado en el paciente. Se consideró que la evidencia era de certeza moderada.

Implicaciones para la investigación: Los investigadores deben adherirse estrictamente a las guías CONSORT para mejorar la calidad del informe sobre las intervenciones. Una descripción minuciosa de los protocolos de intervención es necesaria para asegurar la reproducibilidad.

Cuando se examina la efectividad de los tratamientos psicológicos para los pacientes con epilepsia, el uso de listados estandarizados de CdVRS (QOLIE‐31, QOLIE‐31‐P, y QOLIE‐89) aumentaría la comparabilidad. Por desgracia, existe una laguna crítica en los ECA pediátricos y ECA que incluyen a personas con epilepsia y discapacidades intelectuales.

Finalmente, para aumentar la calidad general de los diseños de estudio de ECA, debe aplicarse una asignación al azar adecuada con ocultación de la asignación y una evaluación de resultados cegada. Debido a que la deserción a menudo es alta en la investigación que requiere la participación activa, debe realizarse un análisis de intención de tratar. También deben evaluarse la fidelidad al tratamiento y la competencia del tratamiento. Estas importantes dimensiones, relacionadas con la evaluación del "riesgo de sesgo", deben ser siempre comunicadas.

PICO

Population

Intervention

Comparison

Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Resumen en términos sencillos

disponible en

Tratamientos psicológicos para los pacientes con epilepsia

Pregunta de la revisión

En esta revisión Cochrane, se quiso averiguar si el bienestar (calidad de vida) de las personas con epilepsia podría mejorarse mediante la participación en terapias psicológicas basadas en habilidades y educativas.

¿Por qué es esto importante?

La epilepsia es una afección cerebral en la que se producen repentinas ráfagas de intensa actividad eléctrica en el cerebro que hacen que los mensajes del cerebro se mezclen, dando lugar a una crisis. Las crisis afectan a las personas de diferentes maneras: pueden causar sensaciones, movimientos o sentimientos inusuales, pérdida de conciencia, caídas, rigidez o convulsiones. Los ataques epilépticos pueden ocurrir repetidamente y sin ningún tipo de desencadenante. Las crisis pueden ocurrir en cualquier momento y en cualquier lugar; pueden aparecer repentinamente y pueden producirse a menudo.

La epilepsia puede deteriorar de forma significativa la calidad de vida y el bienestar de una persona. Por ejemplo, muchas personas con epilepsia experimentan depresión y ansiedad, problemas de memoria, desempleo y discriminación, efectos secundarios adversos de los medicamentos, dificultades para la autonomía y preocupación por las crisis y sus consecuencias.

Los tratamientos para la epilepsia suelen centrarse en detener o reducir el número de crisis que sufre una persona con los mínimos efectos secundarios posibles. Sin embargo, las terapias psicológicas, generalmente administradas por psicólogos, psiquiatras u otros profesionales de la salud, podrían mejorar el bienestar de las personas con epilepsia.

¿Qué se hizo?

El 12 de agosto de 2019, se buscaron en las bases de datos de investigación estudios que midieran los efectos de la terapia psicológica ofrecida a las personas con epilepsia en su calidad de vida. Se encontraron 36 estudios que implicaron a 3526 personas con epilepsia. Los participantes de la mayoría de los estudios fueron adultos (27 estudios); solo tres estudios incluyeron a adolescentes y adultos, y seis a niños y adolescentes.

La mayoría de los estudios (27) midió los efectos de las terapias psicológicas basadas en habilidades. Estos tratamientos enseñan a las personas habilidades que pueden utilizar en su vida diaria. Las técnicas de tratamiento de las terapias basadas en habilidades incluyen: terapia cognitivo‐conductual (TCC), que incluye técnicas de TCC como la respiración, el razonamiento o la visualización; asesoramiento y ejercicios de conciencia plena. Los otros nueve estudios midieron las terapias educativas. Estos tratamientos tenían como objetivo aumentar el conocimiento de las personas sobre la epilepsia y las afecciones relacionadas, los tratamientos para la epilepsia, o sobre cómo funciona el cerebro. Los estudios compararon los efectos de las terapias psicológicas con los cuidados habituales de una persona, medicamentos antidepresivos o el apoyo social.

Los 36 estudios tenían diferentes diseños y evaluaron la calidad de vida de las personas utilizando escalas distintas, por lo que no fue posible compararlos todos. Sin embargo, se pudo comparar los resultados de 11 estudios de terapias basadas en habilidades porque utilizaron la misma escala para evaluar la calidad de vida.

¿Qué se encontró?

Los 11 estudios (con 643 adultos y adolescentes) se realizaron en Europa (tres estudios), EE.UU. (cuatro), Hong Kong (dos), México (uno) y Australia (uno). Las personas de los estudios recibieron un seguimiento y evaluación de 12 semanas a dos años.

Los 11 estudios midieron diferentes aspectos de la calidad de vida. Las personas que recibieron terapias psicológicas basadas en habilidades informaron de una mejor calidad de vida en general comparado con las que recibieron atención habitual (8 estudios), apoyo social (2) o medicamentos antidepresivos (1).

Diez de esos 11 estudios también examinaron subescalas de los cuestionarios de calidad de vida para evaluar aspectos específicos de la calidad de vida. Las personas en estos estudios informaron de mejores resultados en las seis subescalas, ‐ bienestar emocional, energía y cansancio, bienestar general, preocupaciones por las crisis, efectos de la medicación y función social, que se combinan para calcular la calidad de vida general.

Mensajes clave

Se concluyó que las terapias psicológicas basadas en habilidades pueden mejorar el bienestar (calidad de vida) en adultos y adolescentes con epilepsia.

Existe una confianza moderada en el resultado de estos 11 estudios de psicoterapias en personas con epilepsia. Se cree que no es probable que más estudios sobre este desenlace específico de la calidad de vida cambien los hallazgos.

La revisión está actualizada a 12 de agosto de 2019.

Authors' conclusions

Implications for practice

We found moderate‐certainty evidence that skills‐based psychological interventions benefited adults with epilepsy in quality of life. The effect in the intervention groups was significantly better than in the control groups (using usual care, social support, counseling as usual, or selective serotonin reuptake inhibitors). Unfortunately, we found few interventions focused on quality of life in children and adolescents. Only one study (Ring 2018) investigated an intervention in individuals with intellectual disability (IQ < 70). These findings therefore can not be generalized to this population.

Implications for research

Increasing overall quality of reporting

In many cases, the quality of the study design and its implementation appeared to be better than the actual publication suggested. There are mechanisms available to raise the quality of reporting (e.g. submitting manuscripts of RCTs with a CONSORT checklist [Schulz 2010]). Adherence to the CONSORT guidelines and use of the CONSORT checklist may not only increase the effort by study authors and review authors, but it may also raise the quality of reporting in this resource‐intensive field of research. The quality of reporting may be improved with small changes, such as including the descriptor 'assessor‐blinded' in the title, since depending on the study design, this may be the only type of blinding that is feasible in this area of research. Adhering to CONSORT guidelines may be made difficult by a journal's word limitation policies. If that is the case, pertinent details about study design should be submitted as supplementary materials published online. Publication of research designs prior to conducting a study is now required by many journals, which increases transparency. The CONSORT group has developed an extension for trials assessing nonpharmacologic treatments to acknowledge and help navigate the specific challenges that are not addressed in the original CONSORT guidelines (Boutron 2008).

Furthermore, specific information about participant screening and selection will allow clinicians to assess the applicability of an intervention to their clinical setting or to modify an intervention for their patient population. We encourage the reporting of non‐significant study results, since they also make important contributions to the concerns of the whole scientific community.

Increasing comparability by using common and meaningful HRQOL outcome measures

Investigators are encouraged to ensure that their outcome measures match the treatment goal of the investigated intervention (e.g. self‐management, coping, etc.) (Modi 2017); HRQOL may constitute a secondary outcome measure. Despite diverse treatment goals and outcomes, the broad use of the standardized QOLIE inventories would increase comparability of studies investigating psychological treatments for children and adults with epilepsy.

Since many psychological treatments involve patient‐oriented goal setting, it would be interesting to explore if the extended Quality of Life in Epilepsy‐31‐P would provide a more accurate reflection of the treatment effects than the use of the 'non‐personalized' QOLIE‐31, due to the individually‐weighted calculation of scores for the individual's subjective evaluation. This exploration would require the correlation of this extended version with quantitative and qualitative clinical data in trials investigating psychological treatments.

We could not include any pediatric RCTs in the meta‐analysis, due to a lack of epilepsy‐specific HRQOL outcomes. Future pediatric studies should incorporate validated HRQOL measures that qualify as common data elements.

To help with the interpretation and evaluation of clinical rather than statistical relevance of outcomes, we recommend that all studies using HRQOL outcome measures for which a minimum clinically important change has been determined, include the percentage of participants whose results reached a minimum clinically important change, with confidence intervals.

If a minimum clinically important change has not been established, providing effect sizes would help readers to assess the clinical meaning of statistical results.

Increasing overall quality of study designs

In order to increase the overall quality of study designs, adequate randomization and allocation concealment, and blinded outcome assessment should be pursued when conducting RCTs investigating psychological treatments for people with epilepsy. As attrition is often high in this type of research, which requires active participation, an intention‐to‐treat analysis (or other appropriate statistical analysis accounting for follow‐up data) should be carried out, and reasonable power calculations should be conducted to determine appropriate sample sizes. For active and immediate control groups, supportive therapy, social support, or regular counseling can be used as a control group for the effects of attention, which allows blinding of the participants to their treatment arm (Lundgren 2006; Martinović 2006; Tang 2015). To facilitate the attribution of treatment effects, the use of anti‐seizure medications (ASMs) and the resulting changes should be recorded and reported. There are measures available that would allow for the evaluation of treatment fidelity, such as recording and scoring sessions, which should also be used and reported. Besides treatment fidelity, the quality of actual treatment delivery (i.e. the clinical skills of the therapist) should also be actively monitored and reported.

Due to limitations of self‐report measures, future study designs could complement these measures with objective measures, e.g. hospital stays and emergency room visits, sick leaves and injuries, as well as biomarkers, such as functional Magnetic Resonance Imaging (fMRI), quantitative EEG analyses, etc. However, the validity of a number of fMRI studies has recently been questioned, which may have a large impact on the interpretation of weakly‐significant neuroimaging findings (Eklund 2016).

In order to allow for a better assessment of the sustainability of study results, study designs should include a follow‐up assessment of outcome measures at least three months after treatment discontinuation.

Increasing overall generalizability

Impaired intellectual function or disability or both are more common in people with epilepsy than in the general population, especially among people with early‐onset epilepsy. Investigating the applicability of using skills‐based psychological interventions or educational interventions in these populations could therefore provide important clinical insights. There was only one included study (Ring 2018) focused on using a skills‐based psychological intervention for individuals with ID (IQ < 70). We recommend examining this subsample of people with epilepsy as a direction for further research.

Increasing implementation efforts

While an increasing number of clinical trials have investigated the efficacy of psychological interventions for people with epilepsy, only a smaller number of these interventions have outlasted the funding of the clinical trial period, and an even smaller number of interventions have been integrated into usual‐care pathways in locations other than the original study site. Future funded studies may require a section and plan describing how implementation science addresses moving a successful intervention into clinical practice. In order to increase these implementation efforts, it seems essential to a) create training curricula, b) negotiate with insurance companies about acknowledging and reimbursing psychological treatments as part of usual care, and last but not least c) investigate the feasibility of replicating psychological intervention protocols in non‐research settings and in particular in poorly‐resourced environments where most of the world's population resides.

Summary of findings

Open in table viewer

Summary of findings 1. Psychological treatments compared with usual or supportive care

Psychological treatments compared with usual or supportive care
Patient or population: adolescents and adults with epilepsy Setting: outpatient clinic or outpatient clinic and by phone or in‐home sessions and by phone Intervention: skills‐based psychological interventions Comparison: wait‐list control (WLC), usual care (UC) or supportive care (SC) or antidepressant drug treatment
Outcomes	*Comparative effect sizes^ (95% CI)**		№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Wait‐list control, usual care, supportive care or antidepressant drug treatment	Psychological treatments	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
QOLIE‐31 total score^a	The range of mean change in the control groups was −1.9 to 15.96 points.	The range of mean change in the intervention groups was 3.27 to 17.2 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.23 higher (95% CI 3.02 to 7.44 higher) than the control groups	643 (11 RCTs)	⊕⊕⊕⊝ MODERATE^c	2 out of 3 studies that could not be included in meta‐analysis due to use of QOLIE‐89 or QOLIE‐31‐P reported significantly higher postintervention QOLIE total scores in the treatment over the control groups (Hosseini 2016; Yadegary 2015). For narrative synthesis of all other HRQOL outcomes see Table 2.
QOLIE‐31 emotional well‐being subscale^a	The range of mean change in the control groups was −6.23 to 24.95 points.	The range of mean change in the intervention groups was 0.91 to 20.57 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.96 higher (95% CI 0.70 to 9.21 higher) than the control groups	643 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 energy or fatigue subscale^a	The range of mean change in the control groups was −5.3 to 17.69 points.	The range of mean change in the intervention groups was 0.44 to 18.75 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.25 higher (95% CI 1.56 to 8.93 higher) than the control groups	642 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 overall QoL subscale^a	The range of mean change in the control groups was −2.63 to 15 points.	The range of mean change in the intervention groups was 0.13 to 19.64 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.95 higher (95% CI 3.05 to 8.85 higher) than the control groups	639 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 seizure worry subscale^a	The range of mean change in the control groups was −5.18 to 17.26 points.	The range of mean change in the intervention groups was 2.74 to 28.56 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.35 higher (95% CI 1.35 to 7.35 higher) than the control groups	632 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 cognitive functioning subscale^a	The range of mean change in the control groups was −2.71 to 13.17 points.	The range of mean change in the intervention groups was 2.28 to 16.16 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.18 higher (95% CI 1.82 to 6.54 higher) than the control groups	641 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 medication effects subscale^a	The range of mean change in the control groups was −8.11 to 12.04 points.	The range of mean change in the intervention groups was 0.93 to 6.64 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 3.16 higher (95% CI 0.01 to 6.32 higher) than the control groups	643 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 social function subscale^a	The range of mean change in the control groups was −4.28 to 13.98 points.	The range of mean change in the intervention groups was 2.3 to 10.49 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 3.09 higher (95% CI ‐0.17 lower to 6.35 higher) than the control groups	630 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
* Comparative effect sizes were calculated from the mean changes between baseline and post‐intervention in the intervention and control groups. CI: Confidence interval; QOLIE: Quality of life in epilepsy; RCT: randomized controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.
^aRange 0 ‐ 100 points, higher score means higher quality of life. ^bThe median postintervention measurement point was 3 months (8 weeks to 2 years). ^cSerious risk of bias, i.e. included studies share serious risk of performance bias and five included studies share serious risk of attrition bias.

Background

This review is an update of a review previously published in the Cochrane Database of Systematic Reviews (2017, Issue 10; Michaelis 2017).

Description of the condition

Epilepsy is defined as the chronic predisposition of the brain to have recurrent unprovoked seizures. According to the most recent update of the clinically‐oriented definition of epilepsy, the diagnosis can be made after an individual suffers only one reflex or unprovoked seizure, if further diagnostic test results indicate the likelihood of a predisposition to recurring seizures (Fisher 2014). Recent epilepsy definitions also emphasize that epilepsy should not be conceptualized solely in terms of seizures, as many people with epilepsy experience associated behavioral, psychological, and social consequences that form part of their condition (Fisher 2005). It is estimated that between 0.6% (pediatric) to 1% (adult) of the world's population have epilepsy, making it one of the most common neurological conditions (CDC 2012; Russ 2012; WHO 2017).

Aproximately a third to a half of individuals with epilepsy have drug‐resistant seizures (Kwan 2000); even the latest generation of antiseizure medicines (ASMs) have failed to increase the proportion of people with epilepsy who become seizure‐free with drug treatment. Some individuals with drug‐resistant seizures may be eligible for epilepsy surgery (Téllez‐Zenteno 2005). However, in cases where surgical treatment options have been recommended, potential complications of permanent, significant neurological injury and seizure recurrence need to be taken into account (Tanriverdi 2009; Téllez‐Zenteno 2010; Wellmer 2012). Furthermore, the other main epilepsy treatments, surgical resection, neuromodulation (e.g. vagus nerve stimulation, responsive neurostimulation, and deep brain stimulation) and diet, fail to control epileptic seizures fully in 10% to 40% of patients (Jehi 2014).

Individuals with epilepsy often have lower health‐related quality of life (HRQOL) compared to those with other chronic diseases (Wang 2012). Even a single seizure may be associated with reduced HRQOL (Modi 2011). This impact is not surprising, given the potential extent of medical, social, and emotional ramifications of epilepsy. As a result, it is recommended that the comprehensive management of epilepsy should go beyond merely managing seizures, and additionally aim to improve the HRQOL of people with epilepsy (Jacoby 2008).

Factors contributing to poor HRQOL include medical aspects, such as seizure frequency and severity (Camfield 2001; Cramer 1999; Devinsky 1999; Williams 2003), ASM side effects (Benavente‐Aguilar 2004; Gilliam 2004; Loiselle 2016), medication adherence (Wu 2014), as well as socioeconomic status (Loiselle 2016) and psychological comorbidities (Loiselle 2016; Ramsey 2016). Notably, psychological factors, such as distress and loneliness, show robust correlations with HRQOL, while seizure‐related factors appear to be less closely related (Suurmeijer 2001; Baca 2011). When investigating the association between medical parameters and HRQOL, ASM side effects (Loiselle 2016; Modi 2011; Ramsey 2016; Wu 2014), and the number of prescribed ASMs have emerged as stronger predictors of HRQOL than seizure control (Ferro 2013; Ramsey 2016). Individuals with epilepsy are also at increased risk of psychiatric comorbidities or psychological difficulties, which may significantly impact HRQOL (Selassie 2014; Wagner 2015; Scott 2017). For example, children and adolescents with epilepsy are at a three‐ to six‐fold increased risk (21% to 60%) of psychological comorbidities (e.g. attention deficit hyperactivity disorder [ADHD] and depression [Ott 2001; Ott 2003]), compared to the general population and youth with non‐neurological (Davies 2003; Ekinci 2009; Rutter 1970) or neurological medical conditions (Wagner 2015).

Epilepsy and psychiatric disorders share a bi‐directional relationship, which has been supported by both population‐based and experimental studies in human and animal models (Chang 2011; Hesdorffer 2006; Jones 2013; Kanner 2006; Kanner 2009). Individuals with a previous history of psychiatric disorders are four to seven times more likely to develop an unprovoked seizure or chronic epilepsy, compared to individuals without this history (Hesdorffer 2000; Hesdorffer 2006). In a review of seizure incidence in psychopharmacological clinical trials (N = 75,873), the incidence of seizures was significantly lower amongst participants who received antidepressants compared to those receiving placebo (standardized incidence ratio = 0.48; 95% CI 0.36 to 0.61). The study concluded that second‐generation antidepressants, other than bupropion, could potentially have apparent anticonvulsant effects. In addition, the diagnoses of depression, psychotic disorders, and obsessive compulsive disorder were associated with reduced seizure thresholds (Alper 2007). These findings prompted further research into the role of psychological states in the development and manifestation of seizures, as well as the potential effects of psychological therapy on individuals with epilepsy (Kanner 2006; Tang 2014).

Psychiatric comorbidity in epilepsy appears to be strongly associated with psychosocial factors (Gandy 2012). For instance, the increase in depression and anxiety one year after diagnosis of epilepsy is correlated with the degree to which an individual senses loss of self‐control, rather than the actual number of seizures (Velissaris 2012). Moreover, HRQOL is correlated with depression symptoms in epilepsy (Gilliam 2002). Concerns over recurring seizures may diminish HRQOL, even in individuals with well‐controlled epilepsy (Snyder 1990). These concerns may hamper psychosocial functioning and the achievement or maintenance of higher education and employment, despite seizure freedom (Gilbert 2012). By permeating the individual's sense of self‐efficacy and consequently decreasing self‐confidence, concerns about seizure recurrence that stem from the perceived unpredictability of the course of epilepsy can be far more disabling than the seizures themselves (Stevanovic 2007). Reported depressive symptoms were also associated with negative coping, suggesting that interventions targeting negative coping may improve depressive symptoms in youth with epilepsy (Wagner 2010). Daily routines and activities of daily living are often affected, including sleep, work productivity, school, and recreational and sports activities. This impact on work productivity may incur significant indirect costs for the wider economy (Larson 2012; Painter 2014). Notably, the healthcare costs for children with epilepsy in the first year of diagnosis are approximately USD 20,000 per patient. Seizure control, side effects, and HRQOL are strong drivers of healthcare charges (Ryan 2015; Ryan 2016).

Self‐ or family‐management has been identified as a key health variable and is broadly defined as encompassing the personal resources needed to manage a chronic condition in the context of everyday life. Self‐management of adult epilepsy has been defined as “activities that an individual can perform alone that are known to either control frequency of seizures or promote well‐being of the person with seizures” (Dilorio 1992). In pediatric chronic illness, self‐management behaviors are modifiable behaviors linked to influences (e.g. coping responses) through processes (e.g. allocation of treatment responsibility). These self‐management behaviors and processes operate within individual, family, community, and healthcare system domains (Dilorio 1992; Modi 2012; Schilling 2002). Comprehensive evidence in the Institute of Medicine's Report on Epilepsy supports the relevance of self‐management domains in epilepsy, regardless of the age at onset, or of the epilepsy type (Institute of Medicine 2012). An adult self‐management instrument has been developed and published to measure behaviors with psychometrics showing high internal consistency factor reliability (Escoffery 2015a; Escoffery 2015b). Pediatric self‐management instruments are also available, but have tended to focus on specific aspects of self‐management (Smith 2018).

Description of the intervention

In the treatment of epilepsy, physicians aim to reduce seizures using ASMs, surgical interventions (including neuromodulation), or diet treatment. Adjunctive psychological interventions for individuals with epilepsy provided by a range of different professionals (including psychologists, psychiatrists, psychotherapists, nurses and social workers) aim to optimize HRQOL, and to improve mental health and seizure control. Given the high prevalence of mental health disorders in the epilepsy population, and the significant influence epilepsy and its treatments can have on the HRQOL of individuals with epilepsy and their families, psychological interventions are commonly used as an important adjunctive treatment. In addition, psychological treatments may assist with self‐management and adherence to epilepsy management, which is pivotal to improving and maintaining health outcomes for people with epilepsy. To operationalize the definition of 'psychological therapy', studies reviewed included a broad range of interventions that used psychological techniques for children and adults with epilepsy. These interventions can be grouped into two main categories, according to the opinion of the Psychology Task Force of the International League Against Epilepsy (ILAE) and based on general psychotherapy research:

Education‐only interventions provide de‐individualized facts and knowledge to the observational learner, while skills‐based psychological interventions (which may involve educational elements) require an engaged learner who contributes to a personalized curriculum by applying knowledge for behavioral change.

1. Skills‐based psychological interventions

Skills‐based psychological interventions aim to improve HRQOL by improving the person's use of adaptive coping skills. These interventions are usually based on at least one theory of psychotherapy. The specific skills taught can encompass a broad range of treatment methods and treatment goals. Even though different psychotherapy theories use different terminology, therapeutic principles common to many emphasize the development of a person’s awareness of current feelings and repeated patterns of behavior followed by the translation of their enhanced understanding of behavioral patterns into more effective functioning (including interpersonal relationships and processing of emotions) to promote mental health and adjustment to chronic illness. These interventions are designed to increase learning, practice, and generalization of adaptive psychological skills through a variety of psychotherapeutic strategies. However, at the core of skills‐based psychological interventions is the intention to enhance the practice and adoption of adaptive psychological skills in the person's everyday life outside of the intervention session. Skills‐based psychological interventions usually begin with psycho‐educational components to justify the teaching of specific psychological skills. Examples include cognitive behavioral or behaviorally‐based interventions, mindfulness‐based interventions, and other psychotherapeutic methods. The delivery of skills‐based psychological interventions encourages an empathic and supportive approach from the therapist, inviting collaboration with the person. Depending on the treatment goal, certain skills‐based psychological interventions can be focused on self‐ or family‐management interventions or adherence interventions. Self‐ or family‐management interventions typically focus on enhancing skills to improve medication‐taking, managing seizure triggers, avoiding certain foods while on the ketogenic diet, and managing comorbidities associated with epilepsy.

2. Education‐only interventions

Educational interventions (including psychoeducation) are defined as interventions that aim to increase knowledge about epilepsy, its comorbidities, and its treatments, or the working of the brain. They may accommodate the opportunity for participants to learn about certain skills (such as coping skills) but they do not guide participants through their practice, and do not place emphasis on incorporating these skills into their daily living.

How the intervention might work

The high level of psychiatric comorbidities in people with epilepsy has yielded intervention efforts for both children and adults. Several studies, with varying approaches, have been conducted with the objective of improving mental health and HRQOL in adults with epilepsy (e.g. Project UPLIFT, PEARLS [Ciechanowski 2010; Thompson 2010]). For instance, one approach compared a community‐integrated home‐based program (PEARLS) for managing depression in adults with epilepsy with a usual‐care control group. This study found that, over 12 months, the proportion of participants with suicidal ideation differed significantly between groups, increasing by 12% in the control group and decreasing by 24% in the PEARLS group (Ciechanowski 2010). Compared with participants in the control group, participants assigned to the intervention group also had less severe depression.

Another approach is described by Martinović 2006, who delivered an intervention based on cognitive behavioral therapy (CBT) to prevent depressive symptoms in youths with epilepsy who were at risk of depression. Further approaches have attempted to improve well‐being in individuals with epilepsy by using general stress reduction or tolerance techniques, which aim to be effective by reducing psychological stress and its physiological correlates (Novakova 2013). For example, Tang 2015 developed a mindfulness‐based therapy for participants with drug‐resistant epilepsy. Significantly more participants in the mindfulness‐based therapy group had a clinically important improvement in the Quality of Life in Epilepsy Inventory (QOLIE‐31) compared to those in the attention‐placebo intervention control group.

Reiter 2009 developed a multi‐modal therapy for people with epilepsy that included biofeedback, relaxation, aura identification, and behavioral modification. Aura interruption techniques may also be part of psychological therapy, and allow individuals to learn new sets of reactions to pre‐ictal and early ictal phenomena, which may decrease fear of recurring seizures and provide a subjective sense of control (Elsas 2011; Fried 1990; Michaelis 2012). Based on the bi‐directional model of epilepsy and psychological states, psychological therapy for people with epilepsy can also emphasize the individuals’ role and participation in the management of their own condition. Thus, there have also been several programs aimed at improving coping and self‐management of epilepsy in both youths (Wagner 2010) and adults (DiIorio 2011; Gandy 2016). Adherence‐promotion interventions that used intention as a strategy (Brown 2009), or family‐based problem‐solving about adherence barriers (Modi 2013; Modi 2016a), may merit further investigation..

Why it is important to do this review

Psychological treatments have been developed that aim to enhance psychological well‐being and seizure control, and reduce psychiatric comorbidities in people with epilepsy. Establishing evidence of the effects of such interventions is methodologically challenging. A review of the current evidence is needed to help inform future therapeutic recommendations and research designs.

Objectives

To assess the impact of psychological treatments for people with epilepsy on HRQOL outcomes.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials (RCTs), or quasi‐RCTs (e.g. studies in which the randomization is according to the day of the week or date of birth).

Types of participants

Men, women and children of any age with any type of epilepsy, drug‐responsive or drug‐resistant, with or without intellectual disabilities, whether or not they were taking antiseizure medication (ASM).

Types of interventions

For the operational definition of 'psychological treatments', we included a broad range of treatments that were designed to improve health‐related quality of life (HRQOL), seizure frequency and severity, and reduce psychological and psychiatric comorbidities. We explain these different broad types of psychological treatments in detail in the Description of the intervention:

Skills‐based psychological interventions (including educational elements);
Education‐only interventions.

We included studies of comparisons of two or more of the above treatments, and comparisons to 'wait‐list control', 'treatment as usual', and antidepressant pharmacotherapy.

Types of outcome measures

Main outcome measures

We included all studies that reported changes from baseline in validated HRQOL measures. If those studies also reported other quality‐of‐life‐related parameters, symptoms of psychiatric comorbidities or seizure‐related outcome measures we also extracted data from those parameters. We excluded studies without a HRQOL measure.

Primary outcomes

Mean of change from baseline, or comparisons of postintervention scores from validated HRQOL measures

Secondary outcomes

Comparisons of postintervention scores on validated measures of psychiatric comorbidities, such as depressive and anxiety symptoms
Comparisons of postintervention data from validated seizure‐related outcome measures

Search methods for identification of studies

Electronic searches

We ran searches for the original review in March 2016, and subsequent searches in September 2016, February 2019, and August 2019. For the latest update, we searched the following databases on 12 August 2019. There were no language restrictions.

Cochrane Register of Studies (CRS Web); search strategy shown in Appendix 1.
MEDLINE Ovid (1946 to 09 August 2019); search strategy shown in Appendix 2.
PsycINFO EBSCO host (1887 onwards); search strategy shown in Appendix 3.

CRS Web includes randomized or quasi‐randomized controlled trials from the Specialized Registers of Cochrane Review Groups including Epilepsy, the Cochrane Central Register of Controlled Trials (CENTRAL), PubMed, Embase, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (ICTRP).

Searching other resources

References from published studies and relevant systematic reviews

We reviewed the reference lists of retrieved studies and reviews to search for additional reports of relevant studies.

Other sources

We contacted colleagues to ask if they were aware of any studies or unpublished data that we had missed.

Data collection and analysis

Selection of studies

Two review authors (RM and VT) independently assessed trial abstracts for inclusion, resolving disagreements through discussion.

Data extraction and management

The same two review authors independently extracted the following data, using an electronic Cochrane data collection form that we had adapted to fit the scope of this review:

Study methods

Type of intervention used
Design
Dates the study was conducted
Duration of study
Timepoints for outcome assessment
Sequence generation and allocation concealment
Blinding method
Controlled confounding variables
Other 'Risk of bias' concerns
Sources of study funding and potential conflicts of interest

Participants

Total sample size and total number of participants allocated to each group
Age, sex, and gender distribution
Seizure type and epilepsy syndrome
Duration of epilepsy
Etiology of epilepsy
Seizure frequency and severity
Presence or absence of learning disability or intellectual disability (ID)
Presence or absence of psychiatric comorbidity or other medical diagnoses
Antiseizure medication and co‐medication
Setting of the study
Inclusion and exclusion criteria
Country of study

Outcome data

Name and definition of outcome
Units of measurement

Results

Study attrition
Sample size for each outcome
Missing data
Summary data for intervention and control groups (for example, means and standard deviations for all outcomes)

The authors tested the applicability of the data collection form by piloting the form. Again, we resolved any differences of opinion through discussion.

Assessment of risk of bias in included studies

The same two review authors independently assessed risks of bias for each randomized trial using Cochrane's recommended domain‐based evaluation tool for randomized trials, in which we made critical assessments separately for different domains, including selection bias (random sequence generation, allocation concealment), performance bias (blinding of personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias (selective reporting), and other potential sources of bias, including treatment infidelity, treatment competence (in terms of the training background of the professionals who delivered the treatment, and the quality of treatment delivery), and selective recruitment (Higgins 2017).

We examined all outcomes reported in papers for selective outcome reporting. We resolved any differences of opinion by discussion.

Measures of treatment effect

We expressed the treatment effect for each continuous outcome measuring HRQOL as a mean difference (MD) with a 95% confidence interval (CI). For studies that did not provide mean and standard deviation (SD) values for changes from baseline, we used correlation values, baseline values and postintervention values from other studies of comparable intervention method, treatment setting (group versus individual), and total treatment time to estimate change from baseline values.

We performed meta‐analyses only for HRQOL data. Since HRQOL constituted the main outcome measure for this review, and we only included studies that investigated HRQOL, we excluded some studies that covered other outcome measures, e.g. psychiatric symptoms. From this perspective, a meta‐analysis of any outcome other than HRQOL would imply a serious selection bias.

Unit of analysis issues

When assessing randomized trials, we took the level at which randomization occurred into account.

In trials with cluster‐randomization, we considered the biases particular to a cluster‐randomized trial, such as recruitment bias, baseline imbalance, loss of clusters, incorrect analysis, and comparability with individually‐randomized trials, and chose the appropriate measure of analysis (e.g. adjusting results from cluster‐randomized trials with intracluster correlation coefficients to combine with individually‐randomized trials in meta‐analysis if appropriate).

Considering the lasting nature of the intervention in question, a cross‐over trial design would not have been appropriate, because of the likelihood of serious carry‐over. Hence, we would only have included data from the first period.

When assessing multi‐intervention studies, we listed all intervention groups in the Characteristics of included studies tables. We only used the intervention groups relevant to the review in analyses. We would have included studies that included three or more of the interventions listed in Types of interventions as separate comparisons in the analysis.

Dealing with missing data

Whenever possible, we contacted the original investigators to request missing data and clarification of methodology. If we assumed that the data were missing for reasons unrelated to the intervention, we said the data were ‘missing at random’ and we based the analyses on the available data. We discussed the potential reasons for the missing data and addressed the potential impact of missing data on the findings in the Discussion section of the review.

Assessment of heterogeneity

We assessed clinical heterogeneity by examining the distribution of important prognostic variables between studies. To assess the statistical heterogeneity of observed differences of study results, we used the Chi² test, forest plots, and the I² statistic. We judged that an I² greater than 70% and a Chi² result of P < 0.01 to indicate statistical heterogeneity of concern.

Assessment of reporting biases

We compared the reported outcomes with the outcome measures and points of measurements stated in the study methods to assess reporting bias within the publication. We assessed reporting biases by comparing the reported outcomes with the original study protocol. To assess reporting biases and to collect missing information if needed, we contacted all study investigators for their original protocols or comparable documents.

Where we could include 10 or more studies within a meta‐analysis, we visually inspected funnel plots and considered whether any observed asymmetry may be due to publication bias. We produced funnel plots only for total HRQOL scores and not for subscales. As subscales contribute to the total score and publication bias is determined at study level, we determined that if publication bias appeared to be present within analysis of the total scores, it would also be present within the analyses of the subscales.

Data synthesis

To assess whether meta‐analysis was appropriate, we compared the types of interventions and types of outcome measures or scales used in the studies, by tabulating the study characteristics. After we completed this, a group of studies appeared to be sufficiently homogeneous for meta‐analysis (see criteria outlined in Assessment of heterogeneity). We meta‐analyzed the results of clinically and statistically homogeneous studies using Review Manager 5 software (RevMan 2014). We used the inverse variance method for continuous outcomes with a random‐effects model. We conducted a narrative synthesis for any outcomes for which the included studies were not sufficiently homogeneous, or for which we had insufficient data for meta‐analysis.

Subgroup analysis and investigation of heterogeneity

Due to the scope of this review, there were several different interventions of interest, and the included studies were diverse. We identified the following subgroups. However, since they were either comparatively small or the information was unavailable, we did not undertake any subgroup analysis:

Children versus adults
Individuals with drug‐resistant epilepsy versus individuals with drug‐responsive epilepsy
Individuals with primary generalized epilepsy versus individuals with focal epilepsy versus unclassified epilepsy syndromes
Individuals with nocturnal seizures versus individuals with diurnal seizures (seizure‐related outcomes only)
Individuals with seizure warning (aura) versus individuals without seizure warning (seizure‐related outcomes only)
Staff‐based versus non‐staff‐based treatments (i.e. web‐based interventions)
Participants with intellectual disabilities (IQ below 70) would have been separately analyzed

We assessed methodological heterogeneity by examining the study design.

Sensitivity analysis

If reasonable, we would have conducted a sensitivity analysis by comparing the results of a second meta‐analysis, including only studies at low risk of bias, to those in the overall meta‐analysis.

Summary of findings and assessment of the certainty of the evidence

We used the GRADE approach to interpret findings (Schünemann 2017), and GRADEpro GDT software (which imports data from Review Manager 5 software [GRADEPro 2015]), to create a 'Summary of findings' table for the primary outcome of HRQOL.

Results

Description of studies

We searched for RCTs and quasi‐RCTs that investigated psychological treatments for individuals with epilepsy and used validated HRQOL outcome parameters.

Results of the search

The electronic search yielded 2165 titles from the databases outlined, and we found four titles through handsearching. Following the removal of duplicates, 1675 titles remained. We ruled out 1237 titles as irrelevant (i.e. these titles clearly indicated that the studies were neither RCTs nor quasi‐RCTs related to the investigation of psychological interventions for people with epilepsy). We screened the abstracts of the remaining 438 titles for eligibility, and obtained the full texts of 85 reports to assess for eligibility. We excluded 26 full text reports (24 studies, see Figure 1 and Characteristics of excluded studies) because they did not examine HRQOL outcomes. Two studies were ongoing (see Characteristics of ongoing studies).

Figure 1

Study flow diagram.

Included studies

We included 36 completed RCTs (3526 participants) in 57 publications in this review. Table 1 and the Characteristics of included studies tables outline the details of the studies and the components of the interventions. Nine studies were conducted in the USA (Caller 2016; Ciechanowski 2010; DiIorio 2011; Fraser 2015; Gilliam 2019; Pramuka 2007; Sajatovic 2016; Sajatovic 2018; Thompson 2010), six in Germany (Jantzen 2009; May 2002; Meyer 2019; Pfäfflin 2016; Rau 2006; Schröder 2014), three in Iran (Hosseini 2016; Pakpour 2015; Yadegary 2015) and the UK (Dorris 2017; Ridsdale 2018; Ring 2018), two in Australia (Edward 2019; Gandy 2014), Canada (Brown 2019; Hum 2019), Hong Kong (Au 2003; Tang 2015), and Sweden (Lundgren 2006; Lundgren 2008). The remaining studies were conducted in Italy (Beretta 2014), Malaysia (Lua 2013), Mexico (Orjuela‐Rojas 2015), Norway (Helde 2005), Netherlands (Leenen 2018), Serbia (Martinović 2006) and Turkey (Turan Gurhopur 2018).

Open in table viewer

Table 1. Intervention methods, strategies, and treatment goals

	Study (intervention acronym)	Main treatment method	Primary treatment goal	Main treatment strategy	Provider	Setting	Delivery	Timing	Participants
Skills‐based psychological interventions	Au 2003	Cognitive behavioral therapy	Seizure frequency	Stress management, cognitive restructuring, communication skills	Clinical psychologist	Clinic	Group	8 weekly 2‐hour sessions	N = 17 adults with at least 2 seizures per month, with subjectively reported psychological distress
	Ciechanowski 2010 (PEARLS)		Depressive symptoms	Cognitive restructuring to address negative depressive thinking + behavioral activation	Trained social worker	Home‐based + telephone calls	Individual	8 50‐min in‐home sessions in 5 months + 7 monthly 5‐ to 10‐min telephone calls	N = 80 adults with epilepsy with significant depression
	Gandy 2014				Intern psychologist	Clinic	Individual	1 x 1‐ to 2‐hour assessment session + 8 weekly 1‐hour sessions	N = 59 adults with epilepsy
	Gilliam 2019			CBT based on standardized and manual‐based Beck guidelines	Nurse educator and trained lay person with epilepsy	Therapist office	Individual	1‐hour session per week for 16 weeks	N = 98 adults (age 21 ‐ 75) with epilepsy and current major depressive episode
	Hum 2019 (UPLIFT)			see Thompson 2010	Licensed mental health professional and trained layperson with epilepsy	Telephone calls	Group	8 weekly 1‐hour sessions	N = 55 adults with epilepsy and depressive symptoms
	Martinović 2006			Cognitive restructuring to address negative depressive thinking + behavioral activation	NR	Clinic	Group	8 weekly sessions + 4 monthly sessions	N = 32 adolescents with epilepsy and subthreshold depression
	Meyer 2019 (Emyna)			Cognitive restructuring to address negative depressive thinking + behavioral activation	NA	Internet‐based	Individual	5 modules with no fixed sequence, each lasting for 60 ‐ 180 min	N 154 adult (> 18) with active epilepsy and a current diagnosis of moderate depression
	Orjuela‐Rojas 2015				Licensed CBT therapist and psychiatrist	Clinic	Group	12 weekly 90‐min sessions	N = 15 adults with epilepsy and major depression
	Schröder 2014 (Deprexis)				NA	Internet‐based	Individual	9 weekly modules (10 ‐ 60 min)	N = 78 adults with self‐reported depressive symptoms
	Thompson 2010 (UPLIFT)				Master of Public Health student and trained lay person with epilepsy	Internet‐based + telephone calls	Group	8 weekly 1‐hour sessions	N = 53 adults with epilepsy and depression (but not severe depression)
	Dorris 2017	Self‐management program	Quality of life	Medical self‐management and sleep hygiene, coping strategies and problem‐solving techniques based on CBT and mindfulness	Epilepsy nurse and clinical psychologist	Clinic	Group	6 weekly 120‐min sessions	N = 69 children and adolescents aged 12 ‐ 17 with epilepsy
	Fraser 2015 (PACES)		Self‐management	Medical and psychosocial self‐management + epilepsy‐related communication	Psychologist and trained lay person with epilepsy	Clinic	Group	8 weekly 75‐min sessions	N = 83 adults with epilepsy
	Leenen 2018 (ZMILE)		Self‐management and quality of life	Self‐monitoring, risk‐evaluation and management; shared decision‐making, goal‐setting skills	Nurse practitioner	Clinic	Group	5 weekly 2‐hour sessions followed by a 2‐hour booster session after 3 weeks	N = 87 adults with epilepsy and on AEDs
	Sajatovic 2016 (TIME)		Depressive symptoms	Personal goal‐setting exercises (with focus on coping with mental illness and epilepsy), stress management, and training to communicate with care providers	Nurse educator and trained lay person with epilepsy	Clinic	Group	12 weekly 60‐ to 90‐min sessions	N = 35 adults with epilepsy and comorbid mental illness
	Sajatovic 2018		Negative health events	SMART "self‐management for people with epilepsy and a history of negative health events"	Nurse educator and trained lay person with epilepsy	Clinic + telephone intervention calls + telephone maintenance	Group + individual	1 face‐to‐face 60‐ to 90‐min group; 7 Internet‐based group; 6 10‐ to 15‐min telephone maintenance	N = 111 adults with at least 1 negative health event within the past 6 months
	Yadegary 2015		Quality of life	Medical and psychosocial self‐management + seizure communication	NR	Clinic	Group	4 weekly 120‐min sessions	N = 60 adults with epilepsy
	DiIorio 2011 (WebEase)	Motivational interviewing (MI)	Medication adherence + perceived stress	Medication adherence + stress and sleep management	NA	Internet‐based	Individual	3 bi‐weekly modules	N = 194 adults with epilepsy
	Hosseini 2016		Quality of life	Enhancement of internal motivation for change, by overcoming dualism	Psychologist and trained layperson with epilepsy	Clinic	Group	5 sessions in 20 days	N = 56 adults with epilepsy.
	Pakpour 2015		Medication adherence	MI techniques	Health psychologist	Clinic	Individual	3 weekly 40‐ to 60‐min sessions	N = 275 adults with epilepsy
	Lundgren 2006; Lundgren 2008	Mindfulness therapy (MT)	Quality of life	ACT + seizure management	Clinical psychologist	Clinic	Group + individual	5 individual 90‐min sessions + 2 x group 3‐hour sessions + 2 x 1‐hour boosters at 6 and 12 months	N = 27 (Lundgren 2006) N = 18 adults with epilepsy (Lundgren 2008)
	Tang 2015	Mindfulness therapy (MT)	Quality of life	Epilepsy management + mindfulness techniques + seizure‐related acceptance	Clinical psychologist	Clinic	Group	4 x bi‐weekly 2 x.5‐hour sessions	N = 61 adults with drug‐resistant epilepsy
	Brown 2019	Behavior‐change counseling	Physical activity and quality of life	Self‐regulatory skills to support behavior change	Trained research assistant	Clinic	Individual	15‐min sessions: weekly/bi‐weekly/monthly weeks 1 – 4/ 6 – 12/16 – 24	N = Children aged 8 – 14 years with epilepsy
	Caller 2016 (HOBSCOTCH)	Cognitive, memory + self‐management training	Quality of life	Problem‐solving therapy and behavior modification strategies + seizure management + social skills	Specialized nurse	Home‐based + telephone calls	Group + individual	8 weekly 40‐ to 60‐min sessions	N = 66 adolescents and adults with epilepsy and self‐reported memory complaints
	Helde 2005	Epilepsy education + nurse‐led counseling	Quality of life	Personalized counseling + disease knowledge + drug adherence	Specialized nurse	Clinic + phone calls	Group + individual	1‐day group + phone calls every 3 months for 2 yrs	N = 114 adults with epilepsy
	Pramuka 2007	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Psychologist and epilepsy nurse	Clinic	Group	6 weekly 2‐hour sessions	N = 55 adults with epilepsy
	Ring 2018	Learning Disability Epilepsy Specialist Nurse Competency Framework	Seizure frequency and quality of life	Provide care according to guidelines developed by the UK ESNA and UK Royal College of Nursing	Licensed mental health professional and trained lay person with epilepsy	Home visits, telephone, clinics and visits to the local primary care or ID team base	Individual	On an as‐needed basis for 24 weeks	N = 312 adults with epilepsy and intellectual disability
Education‐only interventions	Beretta 2014 (EDU‐COM)	Patient‐tailored medication education	Drug‐related problems	Personalized education on drug interaction and tolerability	Treating physician	Clinic	Individual	1‐hour session + booster session after 1 month	N = 174 adults with epilepsy and chronic comorbidity
	Edward 2019	Epilepsy education program	Seizure frequency	Education program developed based on the self‐determination theory (managing epilepsy and medical care; socializing on a budget, healthy lifestyle, emotional management)	Specialized nurse	Not specified in the publication	Not specified in the publication	1 x 120‐min session	N = 35 adults with epilepsy
	Jantzen 2009 (FLIP&FLAP )	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Trained nurses, social workers, medical doctors or psychologists	Clinic	Group	2‐day course (14 hours)	N = 192 children and adolescents with epilepsy, including parents
	Lua 2013	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	NR	SMS‐based	Individual	11 weekly modules	N = 144 adults with epilepsy
	May 2002 (MOSES)	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Trained nurses, social workers, medical doctors or psychologists	Clinic	Group	2‐day course (14 hours)	N = 383 adolescents and adults with epilepsy
	Pfäfflin 2016	Counseling	Satisfaction with information and support	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Specialized nurse	Clinic	Individual	Delivery during routine visits	N = 187 adults with epilepsy
	Rau 2006 (FAMOSES)	Epilepsy education program	Knowledge + coping	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	NR	Clinic	Group	2‐day course (14 h)	N = 70 children with epilepsy
	Ridsdale 2018 [SMILE (UK)]	Epilepsy education program (May 2002)	Quality of life	see May 2002	Nurse educator and trained lay person with epilepsy	Clinic	Group	2‐day course (16 h)	N = 314 adolescents (≥ 16 years) and adults with poorly‐controlled epilepsy
	Turan Gurhopur 2018	Epilepsy education program	Epilepsy‐specific knowledge, self‐efficacy, quality of life	Modular education program including epilepsy knowledge, seizure management, and social aspects of epilepsy	NR	Clinic	Individual	2 ‐ 3 days with a total of 16 hours	N = 92 including children with epilepsy aged 7 ‐ 18; and parents of children with epilepsy

ACT: acceptance and commitment therapy; AED: anti‐epilepsy drug; CBT: cognitive behavioral therapy; ESNA: EpilepSy Nurses Association; ID: intellectual disability; MI: motivational interviewing

Interventions

The authors grouped the investigated psychological treatments according to the above‐mentioned operational definition of 'psychological treatments' for adults and children with epilepsy (see also Table 1).

1. Skills‐based psychological interventions (27 studies, 2240 participants)

Nine skills‐based psychological interventions were cognitive or behavior‐based interventions, or both, with the primary goal of treating depressive symptoms in adolescents or adults or both, with epilepsy and varying levels of depression severity (Ciechanowski 2010; Gandy 2014; Gilliam 2019; Hum 2019; Martinović 2006; Meyer 2019; Orjuela‐Rojas 2015; Schröder 2014; Thompson 2010). The most common treatment strategies were cognitive restructuring to address depressive thoughts, and behavioral and social activation (see Characteristics of included studies for additional strategies in each study). Five of the studies delivered the intervention on an individual basis (Ciechanowski 2010; Gandy 2014; Gilliam 2019; Meyer 2019; Schröder 2014), while the other four used a group format (Hum 2019; Martinović 2006; Orjuela‐Rojas 2015; Thompson 2010). Four of the interventions were conducted in a clinical setting (Gandy 2014; Gilliam 2019; Martinović 2006; Orjuela‐Rojas 2015), two interventions were Internet‐based (Meyer 2019; Schröder 2014), one intervention was phone‐based (Hum 2019), one was Internet‐based with complementing telephone calls (Thompson 2010), and one was home‐based (Ciechanowski 2010). Three of these interventions included mindfulness techniques (Hum 2019; Schröder 2014; Thompson 2010).

Four studies focused on the primary treatment goal of improving HRQOL. Three of them used mindfulness techniques in combination with seizure management techniques, by introducing acceptance and coping related to seizure disturbances (Lundgren 2006; Lundgren 2008; Tang 2015). Lundgren 2006 and Lundgren 2008 included management of seizure triggers and development of aura interruption techniques. Hosseini 2016 investigated motivational interviewing, which focused on enhancement of internal motivation for coping with epilepsy. One intervention combined a single epilepsy education group session, covering epilepsy knowledge (including the topic of drug adherence) and nurse‐led personalized counseling, with the primary treatment goal of enhancing quality of life (Helde 2005).

Four skills‐based psychological interventions (labeled consumer‐driven psychoeducation by the authors) focused on epilepsy‐specific self‐management behaviors as primary (Fraser 2015; Leenen 2018; Pramuka 2007) or secondary treatment goals (Dorris 2017; primary treatment goal: HRQOL), by discussing medical and psychosocial aspects of epilepsy self‐management in a face‐to‐face group setting with children and adolescents (Dorris 2017) or adults (Fraser 2015; Leenen 2018; Pramuka 2007). Another self‐management intervention applied a similar approach, but evaluated the impact of the intervention on HRQOL outcomes (Yadegary 2015). One Internet‐based self‐management program (WebEase) focused on the primary treatment goals of improving adherence and perceived stress levels, by targeting medication adherence, stress and sleep management (DiIorio 2011). Two consumer‐driven self‐management programs targeted special subgroups: one program was designed for people with comorbid mental illness focused on empowerment and support to increase coping with mental illness and epilepsy (Sajatovic 2016), and one program was designed for people who had recently experienced epilepsy‐related complications and aimed at reduction of such negative health events (Sajatovic 2018). One home‐ and telephone‐based intervention combined self‐management and cognitive training (Home‐Based Self‐management and Cognitive Training Changes lives (HOBSCOTCH) in order to increase quality of life, mood, and objective and subjective neurocognitive functions (Caller 2016).

Two skills‐based psychological interventions focused on seizure control and HRQOL. Au 2003 used CBT‐based components (e.g. cognitive restructuring) with seizure management techniques (e.g. identifying and addressing seizure‐provoking situations) on adults with epilepsy and subjective psychological distress. Ring 2018 investigated a nurse‐led competency framework with the particular focus on supporting adults with epilepsy and intellectual disability with the primary treatment goal of improving seizure frequency and HRQOL

Two studies used motivational interviewing as their primary intervention strategy (Hosseini 2016; Pakpour 2015). Hosseini 2016 applied motivational interviewing with adults with epilepsy in a group format aiming to improve HRQOL; the intervention was designed to enhance internal motivation for changing through exploration, identification, and overcoming doubts and dualism. Pakpour 2015 investigated an adherence intervention using motivational interviewing in an individual setting. In this study, a program was designed to enhance medication adherence behavior and clinical outcomes in people with epilepsy, as measured by drug adherence, drug‐taking behaviors, seizure severity, and HRQOL. An additional study (Brown 2019) used behavioral methods (e.g. performance feedback) to target physical activity in children with epilepsy and positively influence depressive symptoms and HRQOL.

2. Education‐only interventions (9 studies, 1286 participants)

Education‐only interventions focused on epilepsy knowledge, advocacy topics, daily self‐management behaviors, and psychosocial aspects in order to enhance quality of life (six trials: Jantzen 2009; Lua 2013; May 2002; Ridsdale 2018, Turan Gurhopur 2018, Edward 2019), increase knowledge and coping (Rau 2006), or satisfaction of participants with information and support (Pfäfflin 2016), or reduce drug‐related problems (Beretta 2014). Four intervention programs were designed to be delivered in a group setting during a two‐day weekend course (Flip&Flap (Jantzen 2009); FAMOSES (Rau 2006); MOSES (May 2002); SMILE (UK) (Ridsdale 2018). Three of these interventions were geared towards the education of children, adolescents, and their parents (Jantzen 2009, Rau 2006, Turan Gurhopur 2018). One intervention investigated the MOSES material using a short message service (SMS)‐based system to deliver the general content of the educational intervention, complemented by information tailored to the individual (Lua 2013). One intervention provided participant‐tailored medication education in individual sessions in order to reduce drug‐related problems (Beretta 2014). Another provided a brief education intervention on lifestyle self‐management in the control of seizures that was developed based on self‐determination theory in order to improve HRQOL, satisfaction with life and resilience (Edward 2019).

Intervention delivery

A specialized team, usually consisting of medical (doctors, nurses) or mental health specialists (e.g. psychologist, psychiatric nurses, social workers) or both delivered most of the education interventions, except for one educational intervention that was delivered by an epilepsy nurse specialist and an electroencephalography (EEG) technician (Ridsdale 2018). Psychologists with different levels of clinical experience and training delivered most of the skills‐based interventions. Five interventions included a peer coach with epilepsy (Fraser 2015; Hum 2019; Sajatovic 2016; Sajatovic 2018; Thompson 2010). One pragmatic design left the delivery of the educational intervention to the treating physician (Beretta 2014). Skills‐based psychological interventions had a median duration of eight weeks (range 3 weeks to 2 years). Education‐only group interventions (Jantzen 2009; May 2002; Ridsdale 2018; Turan Gurhopur 2018) took two to three days; education‐only interventions with individual sessions comprised one (Edward 2019; Pfäfflin 2016), two (Beretta 2014) or 12 (Lua 2013) sessions. For more detailed information about the duration of each intervention please refer to Characteristics of included studies tables.

Control groups

Fifteen trials included a wait‐list control (WLC) group (Au 2003; DiIorio 2011; Dorris 2017; Fraser 2015; Gandy 2014; Hosseini 2016; Hum 2019; Jantzen 2009; May 2002; Pfäfflin 2016; Rau 2006; Sajatovic 2018; Schröder 2014; Thompson 2010; Turan Gurhopur 2018). Eight studies included an immediate active control group: EpINFO program with education intervention, coping strategies and skill‐building activities (Hum 2019); paper‐based education intervention (Lua 2013); supportive therapy (Lundgren 2006); yoga (Lundgren 2008); counseling as usual (Martinović 2006); pharmacotherapy with a selective serotonin reuptake inhibitor (Gilliam 2019; Orjuela‐Rojas 2015); attention‐placebo social support (Tang 2015). The remaining 13 studies used usual care or treatment as usual as the control group (Beretta 2014; Brown 2019; Caller 2016; Ciechanowski 2010; Helde 2005; Leenen 2018; Meyer 2019; Pakpour 2015; Pramuka 2007; Ridsdale 2018; Ring 2018; Sajatovic 2016; Yadegary 2015). Hum 2019 used both an active control group (epilepsy information and self‐management) and a wait‐list control group. The use of a usual‐care or treatment‐as‐usual design instead of a wait‐list control group was especially comprehensible in long‐term interventions (six months or longer [Ciechanowski 2010; Helde 2005]). One study did not describe in detail the format of the control group used (Edward 2019).

Outcome measures

We organized the outcome measures according to the types of outcome defined in the protocol (Types of outcome measures). The Characteristics of included studies tables outline the outcome measures and measurement points in each study. Altogether, the included studies used more than 50 different outcome measures. Skills‐based psychological interventions carried out a median postintervention follow‐up of six months (range 0 to 18 months). Education‐only interventions carried out a median postintervention follow‐up of six months (range 0 to 12 months).

Health‐related quality of life

'Quality of Life in Epilepsy' inventories (QOLIE)

Twenty‐five studies used the most commonly‐used epilepsy‐specific HRQOL questionnaires to measure outcomes (Quality of Life in Epilepsy‐10/‐31/‐31P/‐48/‐89). Eleven studies used the QOLIE‐31 (Au 2003; Beretta 2014; Caller 2016; Ciechanowski 2010; Fraser 2015; Gandy 2014; Lua 2013; Martinović 2006; Orjuela‐Rojas 2015; Pakpour 2015; Schröder 2014), four studies used the patient‐weighted QOLIE‐P (Leenen 2018; Ridsdale 2018; Tang 2015; Yadegary 2015), four studies used the QOLIE‐89 (Gilliam 2019; Helde 2005; Hosseini 2016; Pramuka 2007), four studies used the QOLIE‐10 (DiIorio 2011; Meyer 2019; Sajatovic 2016; Sajatovic 2018), and single studies used the QOLIE‐48 (Turan Gurhopur 2018) and a subscale item of the QOLIE‐31 to inquire about overall QoL (Pfäfflin 2016). The QOLIE‐31‐P is a modification of the QOLIE‐31, with an additional question about the individual's subjective level of distress in each of the six subscales, which allows for an individually‐weighted calculation of scores for the individual's subjective evaluation (Cramer 2003). All studies using QOLIE‐10, QOLIE‐31, QOLIE‐31‐P, QOLIE‐48 and QOLIE‐89 questionnaires reported pre‐ and postintervention mean scores (± standard deviation [SD]). Three studies included the mean difference between pre‐ and postintervention scores (± SD; [Fraser 2015; Helde 2005; Tang 2015]). Only Tang 2015 reported the percentage of participants achieving a minimum clinically important change.

Of the 25 studies that used the QOLIE measures, we considered 14 skills‐based psychological intervention studies to be sufficiently clinically and methodologically homogeneous for meta‐analysis (Au 2003; Caller 2016; Ciechanowski 2010; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Hosseini 2016; Leenen 2018; Martinović 2006; Orjuela‐Rojas 2015; Pramuka 2007; Tang 2015; Yadegary 2015). We did not include education‐only interventions in the meta‐analysis. Due to substantial baseline differences between intervention and control groups, we used the mean change from baseline (± SD) for the meta‐analysis, rather than postintervention scores (± SD). We sought required data from all authors. Seven study authors provided unpublished data we could include in the meta‐analysis: Caller 2016 provided the unadjusted mean change from baseline (± SD); Fraser 2015 provided the mean change from baseline (± SD) for the control group; Gilliam 2019 and Helde 2005 provided raw data so we could convert the results from QOLIE‐89 to QOLIE‐31; Orjuela‐Rojas 2015 provided raw data to calculate the mean change from baseline (± SD); Tang 2015 and Leenen 2018 provided raw data so we could convert the results from QOLIE‐31‐P to QOLIE‐31. Three studies did not provide the mean change from baseline [± SD; (Au 2003; Ciechanowski 2010; Martinović 2006]). We calculated the mean change from baseline as a difference between pre‐ and postintervention means. In order to calculate an adjusted SD, we grouped these three studies with studies investigating interventions that were comparable in intervention method, treatment setting (group versus individual), and total treatment time: Au 2003 with Tang 2015; Ciechanowski 2010 with Gandy 2014; and Martinović 2006 with Orjuela‐Rojas 2015. This allowed us to calculate the adjusted SD of the mean change from baseline, based on the correlation between pre‐ and postintervention means (± SD) of the studies with which they were grouped. Unfortunately, Martinović 2006 could not provide QOLIE‐31 subscale outcomes, so we only included the total score from his study. We present in narrative form the results of the studies that did not provide the raw QOLIE‐89 data (Hosseini 2016; Pramuka 2007), or raw QOLIE‐31‐P data (Yadegary 2015) that would allow us to convert the results into QOLIE‐31 scores. As a result, the meta‐analysis finally comprised data from 11 studies (Au 2003; Caller 2016; Ciechanowski 2010; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Leenen 2018; Martinović 2006; Orjuela‐Rojas 2015; Tang 2015).

We did not include six studies using QOLIE‐31 outcome measures in the meta‐analysis because of meaningful clinical heterogeneity. In two skills‐based psychological interventions, the intervention delivery was not face‐to‐face; it was either web‐based (Schröder 2014), or SMS‐based (Lua 2013). In one skills‐based psychological intervention, the intervention goal was very narrowly defined: Pakpour 2015: increasing drug‐adherence. Three interventions were 'education only' (Beretta 2014; Pfäfflin 2016; Ridsdale 2018). We present their results in narrative form. Two authors provided raw data that allowed us to calculate and present unpublished QOLIE‐31 scores (Beretta 2014; Schröder 2014).

Other HRQOL outcome measures

Four studies used the World Health Organization Quality of Life instrument, short version (WHOQOL‐BREF); (Hum 2019; Lundgren 2006; Lundgren 2008; Schröder 2014). Four studies used the Satisfaction with Life Scale (SWLS); (Edward 2019; Lundgren 2006; Lundgren 2008; Thompson 2010). Edward 2019 used the Short Form 12 (SF12).

Pediatric studies in particular used heterogeneous HRQOL measures: Brown 2019 used the Childhood Epilepsy Quality of Life scale (CHEQOL) and KIDSCREEN‐27; Dorris 2017 used the Paediatric Quality of Life Inventory PedsQL™ version 4.0 and the Glasgow Epilepsy Outcome Scale for Young Persons (GEOS‐YP); and Rau 2006 used the German questionnaire KINDL.

One study that only included participants with intellectual disability used the Epilepsy and Learning Disabilities Quality of Life (ELDQoL) (Ring 2018).

We present all of these other HRQOL outcomes in narrative form.

Psychiatric comorbidities: depression and anxiety

Several included studies also assessed psychiatric comorbidities. Even though some of them used the same outcome measure, we grouped individual results by outcome measures in narrative form rather than a meta‐analysis. Since HRQOL constituted the main outcome measure of this review, we only included studies that investigated HRQOL, which led to the exclusion of some studies that included psychiatric symptoms as an outcome measure, but not HRQOL. From this perspective, a meta‐analysis of any outcome other than HRQOL would imply a serious selection bias.

Depression

Seventeen studies examined changes in the level of symptoms of depression (Brown 2019; Caller 2016; Ciechanowski 2010; Dorris 2017; Fraser 2015; Gandy 2014; Gilliam 2019; Hum 2019; Leenen 2018; Martinović 2006; May 2002; Orjuela‐Rojas 2015; Ridsdale 2018; Schröder 2014; Sajatovic 2016; Tang 2015; Thompson 2010). Six studies used the Beck Depression Inventory or Beck Depression Inventory‐II (Gilliam 2019; Martinović 2006; Orjuela‐Rojas 2015; Schröder 2014; Tang 2015; Thompson 2010); four studies used the Patient Health Questionnaire (Caller 2016; Fraser 2015; Sajatovic 2016; Thompson 2010). Four studies used the Hospital Anxiety and Depression Scale (Gandy 2014; Leenen 2018; Orjuela‐Rojas 2015; Ridsdale 2018), three other studies used the Neurological Disorders Depression Inventory‐Epilepsy Scale (Caller 2016; Gandy 2014; Hum 2019), Sajatovic 2016 used the Montgomery Asberg Depression Rating Scale as an additional measurement of depression symptom severity, Dorris 2017 used the Paediatric Index of Emotional Distress (PI‐ED), Brown 2019 used the Children's Depression Inventory — Short (CDI‐S), Gilliam 2019 also used the Center for Epidemiologic Studies Depression Scale (CES‐D) and the Mini‐International Neuropsychiatric Interview to assess depression. Two studies assessed suicidal ideation (Ciechanowski 2010; Orjuela‐Rojas 2015). The studies used nine different outcome measures for depression.

Anxiety

Six studies examined changes in the level of anxiety symptoms. Four studies used the Hospital Anxiety and Depression Scale for assessing anxiety (Gandy 2014; Leenen 2018; Orjuela‐Rojas 2015; Ridsdale 2018). One study used the Generalized Anxiety Disorder‐7 (Fraser 2015), and one study used the Beck Anxiety Inventory (Tang 2015).

Seizure‐related outcomes

Thirteen studies measured seizure frequency (Au 2003; Ciechanowski 2010; Gilliam 2019; Jantzen 2009; Leenen 2018; Lundgren 2006; Lundgren 2008; May 2002; Rau 2006; Ridsdale 2018; Ring 2018; Sajatovic 2016; Tang 2015). Four studies used seizure severity measures: the National Hospital Seizure Severity Scale (NHS3) (Leenen 2018), the Liverpool Seizure Severity Scale (Pakpour 2015), the Epilepsy and Learning Disabilities Quality of Life Seizure Severity Scale (ELDQoL‐SSS) (Ring 2018), and the Seizure Severity Index (Tang 2015).

Participants

Most studies evaluated the benefit of interventions for adults with epilepsy. One study investigated CBT for adolescents and young adults with epilepsy (Martinović 2006), Dorris 2017 investigated a consumer‐driven self‐management intervention for children and adolescents, Jantzen 2009, Rau 2006 and Turan Gurhopur 2018 investigated educational interventions for children and adolescents with epilepsy, and Brown 2019 investigated a behavior‐change counseling intervention aimed at increasing physical activity in children with epilepsy. Two studies investigated skills‐based interventions for adolescents and adults (Caller 2016; Helde 2005), and three studies investigated educational interventions for adolescents and adults (May 2002; Pfäfflin 2016; Ridsdale 2018).

Several studies used participants' psychological functioning (e.g. depressive symptoms) as one of the inclusion criteria. Seven studies included only adults with epilepsy and depressive symptoms (Ciechanowski 2010: PHQ‐9 score ≥ 10; Gilliam 2019: CES‐D score > 14; Hum 2019: a minimum score of 12 on the Center for Epidemiologic Studies Depression Scale Revised (CESD‐R); Meyer 2019: at least moderate depression (PHQ > 9); Orjuela‐Rojas 2015: major depression according to the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders‐IV [DSM IV]; Schröder 2014: self‐reported depressive symptoms; Thompson 2010: score of < 38 on the CESD‐R). One study included adults with epilepsy and a DSM IV diagnosis of schizophrenia, schizoaffective disorder, bipolar disorder, or chronic/recurrent major depressive disorder (Sajatovic 2016). Most studies did not specifically recruit participants with comorbid psychiatric diagnoses. Two studies recruited participants who experience elevated levels of psychological difficulties without reaching diagnostic criteria: one study included adolescents and young adults with epilepsy and subthreshold depressive symptoms (Martinović 2006), and one study included adults with epilepsy and self‐reported psychological distress (Au 2003). One study only included adults with epilepsy and other chronic comorbidities, because the intervention targeted adverse effects stemming from drug interactions (Beretta 2014). Another study included adolescents and adults with subjective memory complaints, since the intervention included special cognitive and memory training (Caller 2016).

Ten studies used inclusion criteria related to seizure frequency, epilepsy type, or drug‐responsiveness: Au 2003 only included participants with at least two seizures per month; Brown 2019 and Gilliam 2019 only included participants with at least one seizure in the previous 12 months, Hosseini 2016 only included participants with primary generalized tonic‐clonic epilepsy and uncontrolled seizures; Meyer 2019 only included participants with "active epilepsy" (i.e. having taken ASDs within the past five years or a seizure within the past 10 years), Lundgren 2006 and Lundgren 2008 only included participants with at least four seizures over three months; Ridsdale 2018 only included participants reporting at least two seizures in the previous year; Sajatovic 2018 included participants with at least one negative epilepsy‐related health event within the past year; Tang 2015 only included participants with drug‐resistant epilepsy; and Yadegary 2015 only included participants with at least one seizure during the past year. Altogether, the number of individuals with drug‐responsive epilepsy and primary generalized epilepsy was comparably small in the study populations of all included studies. None of the studies reported whether individuals experienced nocturnal or diurnal seizures, or if individuals experienced focal unaware seizures or prodromal seizure warnings.

Most studies excluded individuals with intellectual disability (ID). Four studies did not explicitly mention ID as an exclusion criterion but intellectual functioning was also not included in the demographic characterization of the patient population (DiIorio 2011; Edward 2019; Hosseini 2016; Yadegary 2015). Only one study investigated a nurse‐led intervention and included only participants with epilepsy and ID (Ring 2018).

We include more details on study participants in Table 1 and the Characteristics of included studies tables. Since the subgroups outlined in the review protocol were either comparatively small or the information was unavailable, we did not undertake any subgroup analysis.

We found two studies that were still ongoing. See Characteristics of ongoing studies for details.

Excluded studies

We excluded 26 RCTs because they did not examine HRQOL outcomes.

Risk of bias in included studies

We include details of our judgments and the rationale in the Characteristics of included studies tables, and display summaries in Figure 2 and Figure 3. We shared the details of our judgments with all study authors prior to the publication of this review, for further clarification.

Figure 2

'Risk of bias' graph: review authors' judgements about each 'risk of bias' domain presented as percentages across all included studies

Figure 3

'Risk of bias' summary: review authors' judgements about each 'risk of bias' domain for each included study

Allocation

Most studies (N = 28) reported an adequate method of random sequence generation (Beretta 2014; Brown 2019; Caller 2016; Ciechanowski 2010; Dorris 2017; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Hum 2019; Leenen 2018; Lua 2013; Lundgren 2006; Lundgren 2008; Martinović 2006; May 2002; Meyer 2019; Pakpour 2015; Pfäfflin 2016; Pramuka 2007; Ridsdale 2018; Ring 2018; Sajatovic 2016; Sajatovic 2018; Schröder 2014; Tang 2015; Turan Gurhopur 2018; Yadegary 2015). Two studies did not provide a sufficient description of the randomization process, hence were classified as unclear (Hosseini 2016; Thompson 2010). Reasons for a high risk of bias rating included quasi‐randomized trial designs, such as a matched design (Au 2003), alternating assignment (DiIorio 2011), the placement of participants in the control group if they were unable to attend the intervention face‐to‐face sessions (Edward 2019), and allocation based on participants' application to one of two available courses in a wait‐list control design (Jantzen 2009; Rau 2006). We rated one study at very serious risk of bias, since the allocation depended on the participants' ability to attend the meetings (Orjuela‐Rojas 2015). One author provided further information to clarify the randomization procedure that had not been sufficiently described in the publication (Rau 2006).

Most of the studies (N = 20) reported proper procedures for allocation concealment (Beretta 2014; Ciechanowski 2010; DiIorio 2011; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Hosseini 2016; Lundgren 2006; Lundgren 2008; Leenen 2018; Martinović 2006; Meyer 2019; Pfäfflin 2016; Ridsdale 2018; Ring 2018; Sajatovic 2016; Schröder 2014; Tang 2015; Turan Gurhopur 2018). Allocation concealment can be considered inherent in study designs investigating a web‐based intervention with a web‐based registration and allocation procedure, hence a study with a low‐quality randomization procedure could indeed feature a high‐quality allocation concealment (DiIorio 2011). Nine studies reported an unconcealed allocation procedure (Au 2003; Caller 2016; Dorris 2017; Jantzen 2009; May 2002; Rau 2006; Sajatovic 2018; Thompson 2010; Orjuela‐Rojas 2015), and seven studies provided insufficient descriptions (Brown 2019; Edward 2019; Hum 2019; Lua 2013; Pakpour 2015; Pramuka 2007; Yadegary 2015). Ten authors provided further information to clarify the allocation concealment procedure that had not been sufficiently described in the publication (Au 2003; Beretta 2014; Dorris 2017; Fraser 2015; Lundgren 2006; Lundgren 2008; May 2002; Ring 2018; Sajatovic 2018; Tang 2015).

Blinding

Blinding of participants and personnel is almost impossible to achieve when studying psychological treatments, so most studies had a high risk of bias (N = 30: Au 2003; Brown 2019; Beretta 2014; Caller 2016; Ciechanowski 2010; DiIorio 2011; Dorris 2017; Fraser 2015; Gandy 2014; Helde 2005; Hosseini 2016; Jantzen 2009; Leenen 2018; Lundgren 2006; Lundgren 2008; May 2002; Meyer 2019; Orjuela‐Rojas 2015; Pakpour 2015; Pfäfflin 2016; Pramuka 2007; Rau 2006; Ridsdale 2018; Sajatovic 2016; Sajatovic 2018; Schröder 2014; Tang 2015; Thompson 2010; Turan Gurhopur 2018; Yadegary 2015). Four studies blinded the participants in both the treatment and the active control group, by telling them that they would participate in an intervention to improve coping with epilepsy (Lua 2013; Lundgren 2006; Lundgren 2008; Tang 2015). This was possible only if the study designs used an immediate and active control arm (social support group in two trials, Lundgren 2006; Tang 2015; yoga in one trial, Lundgren 2008; and paper‐based education material in one trial, Lua 2013). There were no randomized personnel in three studies investigating a web‐based intervention (DiIorio 2011; Meyer 2019; Schröder 2014). Two studies were classified as overall low risk, as the therapists who delivered the treatment (CBT and counseling as usual) were blinded to the participants' group status: the researchers only told the therapists that they would deliver psychological means to improve coping with epilepsy (Martinović 2006). Ring 2018 facilitated blinding of personnel and participants through cluster‐randomization of intervention sites. In one study, the blinding status of the personnel delivering an SMS‐based intervention remained unclear (Lua 2013) and in another study with an active control group and a WLC, the blinding status of personnel and participants remained unclear (Hum 2019).

We considered the risk of non‐blinding of any type of control group that did not receive an immediate control intervention to be especially problematic, since it might lead to baseline imbalances of participant‐reported outcome parameters, due to disappointment, i.e. the impression of having been denied an opportunity. Thus, we rated the risk lower in studies that obtained baseline measures prior to randomization. Thirteen studies reported this procedure (Brown 2019; DiIorio 2011; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Hosseini 2016; Lua 2013; Meyer 2019; Pfäfflin 2016; Ridsdale 2018; Sajatovic 2016; Sajatovic 2018).

Blinding of the assessment of participant‐reported outcome data was adequate in most studies (N = 21: Au 2003; Brown 2019; Beretta 2014; Caller 2016; Ciechanowski 2010; Dorris 2017; Fraser 2015; Helde 2005; Hosseini 2016; Jantzen 2009; Leenen 2018; Martinović 2006; May 2002; Meyer 2019; Orjuela‐Rojas 2015; Pfäfflin 2016; Rau 2006; Ridsdale 2018; Ring 2018; Schröder 2014; Tang 2015). Eight studies provided insufficient information (Brown 2019; Edward 2019; Hum 2019; Lua 2013; Pakpour 2015; Thompson 2010; Turan Gurhopur 2018; Yadegary 2015). Eight studies had a high detection bias, because personnel conducting the outcome assessment were aware of the treatment status (DiIorio 2011; Gandy 2014; Gilliam 2019; Lundgren 2006; Lundgren 2008; Pramuka 2007; Sajatovic 2016; Sajatovic 2018), although additional information was provided by Lundgren 2006 and Lundgren 2008 that outcome assessment was blinded only on seizure‐related data. Thirteen authors provided further information on 16 studies to clarify the blinding of outcome assessment that had been insufficiently described in the publication (Au 2003; Brown 2019; Beretta 2014; DiIorio 2011; Fraser 2015; Gandy 2014; Jantzen 2009; Leenen 2018; Lundgren 2006; Lundgren 2008; Martinović 2006; May 2002; Orjuela‐Rojas 2015; Ring 2018; Sajatovic 2016; Sajatovic 2018).

Incomplete outcome data

We rated three studies at low risk of attrition bias because all randomized participants completed the study (Au 2003; Lundgren 2006; Lundgren 2008). We rated 13 studies at low risk, as there were only a small amount of missing data, which were balanced across the groups, with justifiable reasons (Beretta 2014; Brown 2019; Ciechanowski 2010; Helde 2005; Jantzen 2009; Leenen 2018; Lua 2013; Martinović 2006; Pakpour 2015; Ring 2018; Sajatovic 2018; Tang 2015; Turan Gurhopur 2018). We rated 17 studies at high risk of bias, because of larger amounts of missing data (we applied a cut‐off of 15% for short‐term interventions [less than six months], and 20% for long‐term interventions [at least six months]). Losses were balanced in three studies (Fraser 2015; Gilliam 2019; Ridsdale 2018), and unbalanced in 12 studies (Caller 2016; DiIorio 2011; Dorris 2017; Edward 2019; Gandy 2014; Hosseini 2016; Hum 2019; Meyer 2019; Orjuela‐Rojas 2015; Pfäfflin 2016; Pramuka 2007; Sajatovic 2016). One study excluded participants who had missed more than one intervention session, which indicated that no intention‐to‐treat (ITT) analysis had been undertaken (Hosseini 2016). Three studies with overall high attrition only provided the total number of participants lost to follow‐up, without reporting whether they belonged to the intervention or the control group (May 2002; Rau 2006; Thompson 2010). We assigned an unclear risk to one study that did not provide data on their attrition rate (Yadegary 2015).

The risk of attrition is usually quite high in experimental studies that require regular, active, and personal involvement of study participants, as is the case with psychological treatments. There were two studies that reimbursed their participants for participation in the study, but nonetheless had a high attrition rate (DiIorio 2011; Pramuka 2007).

Selective reporting

We rated 30 studies at low risk of bias as there was no evidence of selective outcome reporting within the publications, when examining all outcomes reported in the papers (Au 2003; Beretta 2014; Brown 2019; Caller 2016; Dorris 2017; Edward 2019; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Hosseini 2016; Hum 2019; Jantzen 2009; Leenen 2018; Lua 2013; Lundgren 2006; Lundgren 2008; May 2002; Meyer 2019; Orjuela‐Rojas 2015; Pakpour 2015; Pfäfflin 2016; Rau 2006; Ridsdale 2018; Ring 2018; Sajatovic 2016; Sajatovic 2018; Tang 2015; Turan Gurhopur 2018; Yadegary 2015). We had initially rated six studies at a high risk of bias, as there was evidence of selective outcome reporting within the publications (Ciechanowski 2010; DiIorio 2011; Martinović 2006; Pramuka 2007; Schröder 2014; Thompson 2010). Two authors provided additional data (DiIorio 2011; Schröder 2014). We therefore ended up assessing four studies at high risk of bias due to evidence of selective outcome reporting within the publications (Ciechanowski 2010; Martinović 2006; Pramuka 2007; Thompson 2010).

Four research groups had previously published their study protocol as a separate publication (Leenen 2018; Meyer 2019; Ridsdale 2018; Sajatovic 2016). We requested study protocols from all other authors. We received 10 responses with complete registered protocols or documentation of the included outcome measures, and assessed the risk against these documents (Beretta 2014; Caller 2016; DiIorio 2011; Fraser 2015; Helde 2005; May 2002; Rau 2006, Sajatovic 2018; Schröder 2014; Tang 2015). We confirmed a rating of low risk of bias in 10 of the studies for which study protocols were available, as there was no evidence of selective outcome reporting following review of the documents (Beretta 2014; Caller 2016; Fraser 2015; Leenen 2018; Meyer 2019; Rau 2006; Ridsdale 2018; Sajatovic 2016; Sajatovic 2018; Tang 2015). For one study, additional outcome measures that would have been part of the scope of this review had originally been planned but then not obtained, due to a change of protocol (Schröder 2014). For three studies, it appeared that additional outcome measures had been obtained during the course of the study, but were not mentioned in the final publication (DiIorio 2011, Helde 2005; May 2002). However, these data were not part of the scope of this review. Nonetheless, we sought and obtained these data in two cases (DiIorio 2011; May 2002). In one case, the omitted outcome data that had aimed at capturing the development of the health economic variables (such as frequency of hospital admissions, missed school or work days, etc.) had been collected, but had never been analyzed (Helde 2005).

Other potential sources of bias

We considered other potential sources of bias, such as language bias, fidelity to the intervention protocol, competence in treatment delivery, and selective recruitment.

Our search yielded only one non‐English publication, which was published in German (Rau 2006). However, we did not search non‐English databases, so language bias remained unclear.

In general, we had little information to judge risks of bias in fidelity to the intervention protocol. Six studies reported the use of measures to assess adherence (Gandy 2014: regular supervision with adherence checklist; Orjuela‐Rojas 2015: regular monitoring with adherence checklist; Ridsdale 2018: all courses were audio‐recorded and, of those, approximately 25% were chosen for the fidelity evaluation; Sajatovic 2016, Sajatovic 2018: sessions were audio‐recorded and non‐interventionist study staff evaluated fidelity; and Gilliam 2019: weekly supervision, monitoring and feedback by lead/senior psychologist with CBT adherence checklist). The results of their analysis were reported in two recent studies (Ridsdale 2018; Sajatovic 2018). Risk of infidelity to treatment protocol was rated low in these latter sudies. As a result, we sought clarification from all authors who had not reported the use of measures to assess fidelity to intervention protocol or the results thereof. Sajatovic 2016 replied that they had a high degree of fidelty and was therefore rated as low risk of bias. Eight studies provided additional details on their attempts to assess fidelity to the intervention protocol, with results (Beretta 2014: monitoring was carried out with regular on‐site monitoring visits and verification of protocol adherence; Dorris 2017: audio recordings of each session were checked for fidelity ratings and regular supervision; Leenen 2018: forms were used to record if all sessions adhered to the intervention and session flip charts were reviewed after sessions) or without results; (Ciechanowski 2010: standard training protocols and supervision; Lundgren 2006 and Lundgren 2008: regular supervision; May 2002: regular supervision and video‐taping of some sessions; Thompson 2010: adherence checklists). Beretta 2014 and Dorris 2017 were therefore rated as low risk of bias. We considered the risk of infidelity to the intervention protocol as low in three studies in which the delivery of the intervention was Internet‐based (DiIorio 2011; Meyer 2019; Schröder 2014). In all other studies risk of infidelity to the intervention protocol was rated unclear (Au 2003; Brown 2019; Caller 2016; Edward 2019; Fraser 2015; Helde 2005; Hosseini 2016; Hum 2019; Jantzen 2009; Martinović 2006; Pakpour 2015; Pfäfflin 2016; Pramuka 2007; Sajatovic 2016; Tang 2015; Turan Gurhopur 2018; Yadegary 2015).

We assessed two dimensions of competence in treatment delivery. First, we reviewed the competence in terms of the professional training background of the personnel who delivered the intervention; in web‐based intervention programs we evaluated the training background of the professionals who had designed the intervention. Second, we reviewed the competence of the measures used to assess the quality of actual treatment delivery. Most studies reported the training background of the personnel delivering the intervention (N = 21: Au 2003; Beretta 2014; Caller 2016; Ciechanowski 2010; DiIorio 2011; Dorris 2017; Edward 2019; Gandy 2014; Lundgren 2006; Lundgren 2008; Helde 2005; Hosseini 2016; Leenen 2018; Meyer 2019; Orjuela‐Rojas 2015; Pakpour 2015; Pfäfflin 2016; Ring 2018; Ridsdale 2018; Sajatovic 2016; Sajatovic 2018). Seven studies provided additional details for training background (Brown 2019; Fraser 2015; Gilliam 2019; May 2002; Schröder 2014; Tang 2015; Thompson 2010), We rated the risk of bias as low for this domain in these studies. Four studies reported the use of measures to assess competence (Pakpour 2015: assessment of empathy, use of open‐ended questions, etc; Ridsdale 2018: all courses were audio‐recorded and, of those, approximately 25% were chosen for the competence evaluation; Sajatovic 2016: sessions were audio‐recorded and non‐interventionist study staff evaluated competence; Thompson 2010: supervision) The results of their analysis were reported in three studies (Pakpour 2015; Sajatovic 2016; Ridsdale 2018). We rated the risk of bias as low for competence in intervention delivery in these studies that reported results of the assessment. We sought clarification from all authors who had not reported the use of measures to assess competence or the results thereof. Six authors provided additional details about their attempts to assess competence, with results (Gilliam 2019: session recordings were reviewed for clinical supervision and quality assurance; Lundgren 2006 and Lundgren 2008: regular supervision) or without results (Gandy 2014: no measures to assess competence were used; May 2002: regular supervision and video‐taping of some sessions; Pfäfflin 2016: regular supervision; Sajatovic 2018: sessions were assessed qualitatively for competence [rapport and empathy, engagement, timing, etc.]). As a result, we judged risk of incompetence to be low in three studies (Gilliam 2019; Lundgren 2006; Lundgren 2008), and competence to deliver the intervention protocol to be unclear in all other studies.

In general, we had little information to judge risks of bias in selective recruitment. Risk of selective recruitment was considered low in 12 studies that reported a consecutive recruitment procedure at the intervention site including screening of all patients for eligibility (Beretta 2014; Caller 2016; Dorris 2017; Gilliam 2019; Helde 2005; Pakpour 2015; Pfäfflin 2016; Ridsdale 2018; Ring 2018; Sajatovic 2016; Sajatovic 2018; Tang 2015; Thompson 2010). Risk of selective recruitment was considered high in all studies whose recruitment procedure involved subjective criteria (N = 2: Au 2003: subjective report of psychological distress; Leenen 2018: excluded patients who were not able or willing to function in group activities based on clinical judgment) or advertisement (N = 9: DiIorio 2011; Fraser 2015; Gandy 2014; Lundgren 2006; Lundgren 2008; May 2002; Meyer 2019; Rau 2006; Schröder 2014) or convenience sampling (N = 1: Hosseini 2016). Recruitment procedures were unclear in 11 studies (Brown 2019; Ciechanowski 2010; Edward 2019; Hum 2019; Jantzen 2009; Lua 2013; Martinović 2006; Orjuela‐Rojas 2015; Pramuka 2007; Turan Gurhopur 2018; Yadegary 2015)

Effects of interventions

See: Summary of findings 1 Psychological treatments compared with usual or supportive care

We have listed our outcomes in Table 2, organized in alphabetical order and categorized according to our operational definition of psychological treatment types.

Open in table viewer

Table 2. Effects of interventions

	Study (intervention acronym)	HRQOL	Depression	Anxiety	Seizure‐related outcomes	Additional outcomes	Time points measured
Skills‐based psychological interventions	Au 2003	QOLIE‐31^b	NA	NA	seizure frequency^a,d	ESES	1) baseline 2) postintervention
	Brown 2019	CHEQOL^a,d, KIDSCREEN‐27^a,d	CDI‐S^a,d	NA	NA	Physical activity^a	1) baseline 2) postintervention (28 after baseline) 3) 52‐week follow‐up
	Caller 2016 (HOBSCOTCH)	QOLIE‐31^a,b	PHQ‐9^d NDDI‐E^d	NA	NA	Self‐reported cognitive and executive function	1) baseline 2) postintervention
	Ciechanowski 2010 (PEARLS)	QOLIE‐31^b	HSCL‐20^a,d suicidal ideation^a,c	NA	seizure frequency^d	NA	1) baseline 2) postintervention 3) 12‐month follow‐up 4) 18‐month follow‐up
	DiIorio 2011 (WebEase)	QOLIE‐10^d	NA	NA	NA	ESI‐R^a, ESMS^a, MAS^a, PSQI^a, PSS^a, Epilepsy Knowledge Profile	1) baseline 2) postintervention 3) 12‐week follow‐up
	Dorris 2017	PedsQL^a,d GEOS‐YP^a,d	PI‐ED^d	NA	NA	EKP‐G, SSEC‐C, B‐IPQ	1) baseline 2) postintervention 3) 3‐month follow‐up 4) 6‐month follow‐up
	Fraser 2015 (PACES)	QOLIE‐31^b	PHQ‐9^c	GAD‐7^d	NA	ESES^a, ESMS^a	1) baseline 2) postintervention 3) 6‐months follow‐up
	Gandy 2014	QOLIE‐31^b	HADS‐D^a,c NDDI‐E^a,c	HADS‐A^a,d	NA	NA	1) baseline 2) postintervention 3) 3‐month follow‐up
	Gilliam 2019	QOLIE‐89^b	BDI II^{a, d} CES‐D^{a, d}	NA	Focal impaired seizures/month^d , Generalized tonic‐clonic seizures/month^d	Adverse events profile^d	1) baseline 2) 8‐week interim assessment 3) 16‐week (postintervention)
	Helde 2005	QOLIE‐89^a.b	NA	NA	NA	VAS scale	1) baseline 2) postintervention
	Hosseini 2016	QOLIE‐89^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
	Hum 2019 (UPLIFT)	WHOQOL‐BREF^a,d	QIDS^a,c NDDIE^a,d	NA	NA	NA	1) baseline, 2) 6‐month follow‐up 3) 12‐month follow‐up
	Leenen 2018 (ZMILE)	QOLIE‐31‐P^b,d	HADS‐D^d	HADS‐A^d	NHS3^d	GSES^a, GSES^a, MEMS, MARS, UPCC, SIDAED	1) baseline 2) 3‐month follow‐up 3) 6‐month follow‐up
	Lundgren 2006	SWLS^a WHOQOL‐BREF^a	NA	NA	seizure frequency^c seizure index^c	NA	1) baseline 2) postintervention 3) 6‐month follow‐up 4) 12‐month follow‐up
	Lundgren 2008	SWLS^a, WHOQOL‐BREF^a	NA	NA	seizure frequency^d, seizure index^c	NA	1) baseline 2) postintervention 3) 6‐month follow‐up 4) 12‐month follow‐up
	Martinović 2006	QOLIE‐31^b	BDI^a,d CES‐D^a,d HAMD^a,c	NA	NA	NA	1) baseline 2) postintervention 3) 9‐month follow‐up
	Meyer 2019	QOLIE‐10^c	PHQ‐9^{a, c}, NDDIE^c DASS^c WSAS^c	GAD‐7^c	NA	NA	1) baseline 2) 3‐month (postintervention) 3) 6‐month follow‐up 4) 9‐month follow‐up
	Orjuela‐Rojas 2015	QOLIE‐31^b	BDI^a,c HADS‐D^a,d MINI^a,d	HADS‐A^a,d	NA	NA	1) baseline 2) mid‐intervention 3) postintervention
	Pakpour 2015	QOLIE‐31^c	NA	NA	LSSS^d	MARS^a; for additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention 3) 6‐month follow‐up
	Pramuka 2007	QOLIE‐89^a,d	NA	NA	NA	ESES, WPSI, LOC	1) baseline 2) postintervention 3) 6‐month follow‐up
	Ring 2018	ELDQoL‐SSS^d	NA	NA	ELDQoL‐SSS^a,d	For additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Sajatovic 2016 (TIME)	QOLIE‐10^d	MADRS^a,d PHQ‐9^d	NA	seizure frequency^d	BPRS, GAF, WHODASII, PSQI, ESES, MSPSS, ISMI, ESS	1) baseline 2) postintervention 3) 16 week‐follow up
	Sajatovic 2018	QOLIE‐10^c SF‐36 PCS^c SF‐36 MCS^c	PHQ‐9^c MADRS^c	NA	LSSS^c	ESES^c, MSPSS^d, ESMS^c, ESS^{d, NHEa}	1) baseline 2) 10‐week follow‐up 3) 24‐week follow‐up
	Schröder 2014 (Deprexis)	QOLIE‐31^d WHOQOLBREF^d	BDI^a,c	NA	NA	NA	1) baseline 2) postintervention
	Tang 2015	QOLIE‐31‐P^a,b	BDI‐II^c	BAI^c	seizure frequency^c SSI^d	NA	1) baseline 2) postintervention
	Thompson 2010 (UPLIFT)	SWLS^d	BDI^a,c	NA	NA	DCSES, SCS, knowledge and skills, BRFSS	1) baseline 2) postintervention 3) 4‐month follow‐up
	Yadegary 2015	QOLIE‐31‐P^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
Education‐only interventions	Beretta 2014 (EDU‐COM)	QOLIE‐31^d	NA	NA	NA	Drug‐related problems^a	1) baseline 2) postintervention 3) 6‐month follow‐up
	Edward 2019	SF12‐PCS (ns) SF12‐MCS (ns) SWLS (ns)	NA	NA	NA	CD‐RISC (ns), MMAS‐8 (ns)	1) baseline 2) 6‐month follow‐up postintervention
	Jantzen 2009 (FLIP&FLAP )	DISABKIDS^c	NA	NA	Seizure‐free episode^d	Disclosure of epilepsy	1) baseline 2) postintervention
	Lua 2013	MQOLIE‐30^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
	May 2002 (MOSES)	SF‐36^a,d	Depression Scale D‐S^a,d	NA	Seizure frequency^c	For additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Pfäfflin 2016	QOLIE‐31 (Overall QoL)^d	HADS‐D^d	HADS‐A^d	NA	Satisfaction with information and support^a	1) baseline 2) postintervention
	Rau 2006 (FAMOSES)	KINDL^d	NA	NA	Seizure frequency^d	Epilepsy knowledge^a; for additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Ridsdale 2018 (SMILE [UK])	QOLIE‐31‐P^a,d	HADS‐A^d	HADS‐A^d	seizure frequency and seizure recency^d	Impact of epilepsy, stigma of epilepsy, medication adherence, self‐mastery and control	1) baseline 2) 6‐month follow‐up 3) 12‐month follow‐up
	Turan Gurhopur 2018	QOLIE‐48^c	NA	NA	NA	SSES‐C^c, EKTC^c	1) baseline (pretest), 2) immediately after intervention (post‐test 1), 3) 1‐month follow‐up 4) 3‐month follow‐up

^aprimary outcome measure(s) in study.
^bincluded in meta‐analysis.

Interpretation of post‐intervention outcomes

^cSignificant improvement in treatment group when comparing post‐intervention outcomes of treatment and control group.
^dNo significant difference between treatment and control group at postintervention based on mean comparisons.

NA: not applicable
ns ‐ Not specified, with no information in the publication suggesting significant difference between treatment and control group at post‐intervention based on mean comparisons

Outcome Measures

BAI ‐ Beck Anxiety Inventory; BDI or BDI II ‐ Beck Depression Inventory or Beck Depression Inventory II; B‐IPQ ‐ Brief ‐ Illness Representations Questionnaire; BPRS ‐ Brief Psychiatric Rating Scale; BRFSS ‐ Behavioral Risk Factor Surveillance System; CES‐D ‐ Center for Epidemiological Study on Depression scale; CDI‐S ‐ Children's Depression Inventory ‐ Short; CD‐RISC ‐ Connor‐Davidson Resilience Scale; CHEQOL ‐ Childhood Epilepsy Quality of Life scale; DASS21 ‐ Depression Anxiety Stress Scale‐21; DISABKIDS ‐ Modular HRQOL questionnaire; DCSES ‐ Depression Coping Self‐Efficacy Scale; EKP‐G ‐ Epilepsy Knowledge Profile‐General; EKTC ‐ Epilepsy Knowledge Test for Children; ELDQoL‐SSS ‐ Epilepsy and Learning Disabilities Quality of Life Seizure Severity Scale; ESES ‐ Epilepsy Self‐Efficacy Scale; ESMS ‐ Epilepsy Self‐Management Scale; ESS ‐ 10‐item Epilepsy Stigma Scale; ESI‐R ‐ Revised Epilepsy Stressor Inventory; GAD‐7 ‐ Generalized Anxiety Disorder‐7; GAF ‐ Global Assessment of Functioning; GEOS‐YP ‐ Glasgow Epilepsy Outcome Scale for Young Persons; GSES ‐ General Self‐Efficacy Scale; HADS ‐ Hospital Anxiety Depression Scale; HAMD ‐ Hamilton Depression Scale; ISMI ‐ Internalized Stigma of Mental Illness Scale; HSCL‐20 ‐ Hopkins Symptom Checklist‐20; KINDL ‐ Gesundheitsbezogene Lebensqualität und psychosoziale Auswirkungen der Epilepsie (Health‐related Quality of Life and psychosocial consequences of epilepsy); LOC ‐ Locus of Control Scale; LSSS ‐ Liverpool Seizure Severity Scale; MADRS ‐ Montgomery‐Asbery Depression Rating Scale; MARS ‐ Medication Adherence Report Scale; MAS ‐ Medication Adherence Scale; MEMS ‐ Medication Event Monitoring System; mBDI ‐ Modified Beck Depression Inventory; MCMI‐III ‐ Millon Clinical Multiaxial Inventory‐III; MINI ‐ Mini International Neuropsychiatric Interview; MMAS‐8 ‐ Morisky Medication Adherence Scale (MMAS‐8); MQOLIE‐30 ‐ Malay Quality of Life Inventory in Epilepsy‐30; MSPSS ‐ Multidimensional Scale of Perceived Social Support; NDDI‐E ‐ Neurological Depressive Disorders Inventory‐Epilepsy; NHS3 ‐ National Hospital Seizure Severity Scale; PedsQL ‐ Paediatric Quality of Life Inventory PedsQL™; PHQ‐9 ‐ Patient Health Questionnaire‐9; PI‐ED ‐ Paediatric Index of Emotional Distress; PSQI ‐ Pittsburgh Sleep Quality Index; PSS ‐ Perceived Stress Scale; QOLIE‐31, QOLIE‐31‐P, QOLIE‐89 ‐ Quality of Life in Epilepsy‐31, Patient‐weighted Quality of Life in Epilepsy‐31, Quality of Life in Epilepsy‐89; SCS ‐ Self‐compassion Scale; SF12‐PCS ‐ Short‐Form 12 Physical Health Score; SF12‐MCS ‐ Short Form 12 Mental Health Score; SF‐36 ‐ Short‐Form 36; (PCS ‐ physical health score, MCS ‐ mental health score); SIDAED ‐ Side‐effects of Antiepileptic Drugs; SSEC‐C ‐ Seizure Self Efficacy Scale for Children; SSI ‐ Seizure Severity Index; SWLS ‐ Satisfaction with Life Scale; UPCC ‐ Utrecht Proactive Coping Competence; VAS scale (Helde 2005) ‐ General satisfaction with the follow‐up by the Neurological Clinic during the last 2 years; WHODASII ‐ World Health Organization Disability Assessment Schedule II; WHOQOL‐BREF ‐ World Health Organization Quality of Life instrument, short version; WPSI ‐ Washington Psychosocial Seizure Inventory; WSAS ‐ Work and Social Adjustment Scale; 4‐point Likert scale (Martinović 2006) ‐ Rating of positive and negative thoughts

Health‐related quality of life

Quality of Life in Epilepsy Inventory (QOLIE‐10, QOLIE‐31, QOLIE‐31‐P, QOLIE‐48, QOLIE‐89)

Eleven studies (643 participants) contributed data to the meta‐analysis; they investigated comparable skill‐based psychological interventions and used the most common quality‐of‐life tool (QOLIE‐31/‐31‐P/‐89) as their outcome measure (also see summary of findings Table 1). Seven studies used QOLIE‐31 (Au 2003; Caller 2016; Ciechanowski 2010; Fraser 2015; Gandy 2014; Martinović 2006; Orjuela‐Rojas 2015). Two studies used QOLIE‐31‐P (Leenen 2018; Tang 2015) and two studies used QOLIE‐89 (Gilliam 2019; Helde 2005); all four study authors provided raw data allowing transformation to QOLIE‐31 for the meta‐analysis.

Meta analysis results

Among the 11 studies, two studies were of adolescents and adults (Caller 2016; Helde 2005), one of adolescents and young adults (Martinović 2006), and eight of adults (Au 2003; Ciechanowski 2010; Fraser 2015; Gandy 2014; Gilliam 2019; Leenen 2018; Orjuela‐Rojas 2015; Tang 2015). We found statistically significant mean changes in the total score and each subscale measure, except social function. A positive mean change indicated a postintervention improvement.

Total score (11 RCTs, 643 participants): significant mean change of 5.23 points (95% CI 3.02 to 7.44; P < 0.001; Chi² P = 0.08; I² = 41%); Analysis 1.1; Figure 4
Overall QoL (10 RCTs, 639 participants): significant mean change of 5.95 points (95% CI 3.05 to 8.85; P = < 0.001; Chi² P = 0.11; I² = 37%); Analysis 1.2; Figure 5.
Energy and fatigue (10 RCTs, 642 participants): significant mean change of 5.25 points (95% CI 1.56 to 8.93; P = 0.005; Chi² P = 0.05; I² = 48%); Analysis 1.3.
Emotional well‐being (10 RCTs, 643 participants): significant mean change of 4.96 points (95% CI 0.70 to 9.21; P = 0.002; Chi² P = 0.002; I² = 66%); Analysis 1.4.
Seizure worry (10 RCTs, 632 participants): significant mean change of 4.35 points (95% CI 1.35 to 7.35; P = 0.005; Chi² P = 0.42; I² = 2%); Analysis 1.5.
Cognitive functioning (10 RCTs, 641 participants): significant mean change of 4.18 points (95% CI 1.82 to 6.54; P < 0.001; Chi² P = 0.69; I² = 0%); Analysis 1.6.
Medication effects (10 RCTs, 643 participants): significant mean change of 3.16 points (95% CI 0.01 to 6.32; P = 0.05; Chi² P = 0.62; I² = 0%); Analysis 1.7.
Social function (10 RCTs, 630 participants): non‐significant mean change of 3.09 points (95% CI −0.17 to 6.35; P = 0.06; Chi² P = 0.68; I² = 0%); Analysis 1.8.

Figure 4

Forest plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.1 QOLIE‐31‐ total score.

Figure 5

Forest plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.2 QOLIE‐31 ‐ overall QoL subscale.

Funnel plots for analyses of QOLIE‐31 total score and overall QoL are provided in Figure 6 and Figure 7, respectively. Within both plots, the Orjuela‐Rojas 2015 study is clearly distinguishable from the other studies, but we interpret that this visually different result is due to this study having the smallest sample size and the most imprecise results (largest standard error) of all included studies, and that there is no clear evidence of publication bias from visual inspection of asymmetry of the plots.

Figure 6

Funnel plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.1 QOLIE‐31‐ total score.

Figure 7

Funnel plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.2 QOLIE‐31 ‐ overall QoL subscale.

Narrative results

We report the results of the 14 studies that used QOLIE‐inventories and could not be included in the meta‐analysis in narrative form. They were excluded because they were not skills‐based but education‐only interventions (four studies: Pfäfflin 2016, Ridsdale 2018; Beretta 2014, Lua 2013), or the uniqueness of the intervention protocol (two studies: Meyer 2019; Pakpour 2015), or inadequate data (DiIorio 2011; Hosseini 2016; Meyer 2019; Pramuka 2007; Sajatovic 2016; Sajatovic 2018; Turan Gurhopur 2018; Yadegary 2015).

Seven studies reported significant improvements in the treatment group when comparing the mean postintervention outcomes:

Hosseini 2016 reported a significant mean increase in the total score for the intervention group (mean change 35.95 [SD 8.74]; P < 0.001) and a significant mean decrease for the control group (mean change −8.07 [SD 8.91]; P < 0.001);
Lua 2013 (intervention mean 69.2 [SD 17.4] versus control mean 58.4 [SD 13.6]; P = 0.007);
Meyer 2019 (intervention mean 32.50 [SD 5.12] verses control mean 30.91 [SD 5.05]; P < 0.01 using intension‐to‐treat analysis);
Pakpour 2015 (intervention mean 62.14 [SD 13.21] versus control mean 56.01 [SD 12.12]; P < 0.001 [adjusted for variables such as age and gender]);
Sajatovic 2018 (SMART intervention mean 2.52 [SD 0.9] versus wait‐list control mean 2.99 [SD 08] in which higher scores indicated worse quality of life; P < 0.001);
Turan Gurhopur 2018 (intervention mean increase 2.54 ± 0.238 versus control mean increase 2.26 ± 0.254, P < 0.002);
Yadegary 2015 (intervention mean 72.18 [SD 11.34] versus control mean 53.49 [SD 15.97]; P < 0.001).

The remaining seven studies did not report a significant mean postintervention difference in total scores between the treatment and control groups:

Beretta 2014 (intervention mean 63.00 [SD 15.48] versus control mean 65.04 [SD 14.38]; P value was reported to be non‐significant, without precise value);
DiIorio 2011 (intervention mean 33.77 [SD 7.96] versus wait‐list control mean 33.27 [SD 7.52]; P = 0.731);
Pfäfflin 2016 (intervention mean 68 (SD 21) versus control mean 66 [SD 20], P value was reported to be non‐significant, without precise value);
Pramuka 2007 (intervention mean 67.3 [SD 2.6] versus control mean 65.0 [SD 2.8]; P value was reported to be non‐significant, without precise value);
Ridsdale 2018 (intervention mean 66.3 [SD 13.0] versus control mean 65.5 [SD 14.0]; P = 0.195);
Sajatovic 2016 (intervention mean 2.38 [SD 0.60] versus control mean 3.03 [SD 0,79]; P = 0.129);
Schröder 2014 (intervention mean 50.55 [SD 3.69] versus control mean 52.22 [SD 3.19]; P = 0.667).

The potential impact of the three studies that did not contribute data to the meta‐analysis was probably small (Hosseini 2016 [28 participants]; Yadegary 2015 [30 participants]; Pramuka 2007 [31 participants]), especially since two studies also reported significantly higher postintervention HRQOL in the treatment over the control groups (Hosseini 2016; Yadegary 2015).

Since most of the included studies had at least some bias issues, we did not perform a sensitivity analysis to compare studies at low risk of bias with studies at high risk of bias.

Other HRQOL outcome measures

Thirteen studies used other HRQOL outcome measures (Brown 2019; DiIorio 2011; Dorris 2017; Edward 2019; Hum 2019; Jantzen 2009; Lundgren 2006; Lundgren 2008; May 2002; Rau 2006; Ring 2018; Schröder 2014; Thompson 2010). We report the results in narrative form.

The World Health Organization Quality of Life instruction, short version (WHOQOL‐BREF) was used in four studies (Hum 2019; Lundgren 2006; Lundgren 2008; Schröder 2014). Lundgren 2006 and Schröder 2014 reported a non‐significant mean difference between groups (Lundgren 2006: intervention mean 58.36 [SD 9.66] and the control mean 55.31 [SD 6.59], P value was non‐significant without precise value reported; Schröder 2014: intervention mean 75.9 [SD 15.04] versus control mean 78.62 [SD 17.39], P value was non‐significant without precise value reported). Hum 2019 and Lundgren 2008 reported significant improvement in the intervention group (Hum 2019: P = 0.019; Lundgren 2008: P < 0.01) but not in the control group; however, the significance level for a group difference at postintervention was not specified (Hum 2019: mean changes from baseline scores for the intervention group, active control group, and wait‐list control group were 6.88 [SD 2.6], 5.03 [SD 2.6], and 0.76 [SD 4.2], respectively; Lundgren 2008: postintervention group mean 57.2 [SD 7.2] versus control group mean 60.2 [SD 8.6]).

The Satisfaction with Life Scale (SWLS) was used by four studies (Edward 2019; Lundgren 2006; Lundgren 2008; Thompson 2010). Edward 2019 reported the mean scores at postintervention of the intervention group (mean 24.4 [SD 7.83] versus the control group 25.3 [SD 7.44]) without mentioning the significance level. Lundgren 2006 reported a significant group difference at postintervention (intervention group mean 23.28 [SD 4.58] versus control group mean 13.85 (SD 5.98); P < 0.05). Lundgren 2008 did not specify the significance level of the group difference at postintervention (intervention group mean 21.8 [SD 6.3] versus control group mean 21 [SD 7.1]). Thompson 2010 reported a non‐significant difference between groups (21 [treatment mean] versus 18 [wait‐list control mean]; P = 0.090; SD was not reported).

Seven studies used different HRQOL outcome measures:

Brown 2019 used the Childhood Epilepsy Quality of Life scale (CHEQOL: intervention mean of 77.5 [SD 13.3] versus control mean of 78.9 [SD 12.4]) and the KIDSCREEN‐27 (intervention mean 50.4 [SD 10] versus control mean 47.5 [SD 7.6], but the significance level was not significant);
Dorris 2017 used two measures for quality of life, both showed non‐significant postintervention mean between the intervention and control groups; the Paediatric Quality of Life Inventory PedsQL version 4.0 (intervention mean of 67.61 [SD 14.10] versus control mean of 66.93 [SD 17.28]) and the Glasgow Epilepsy Outcome Scale for Young Persons (GEOS‐YP) (intervention mean 63.82 [SD 14.43] versus control mean 66.83 [SD 11.85]);
Edward 2019 reported the postintervention mean of the SF‐12 physical health score (PCS) (intervention group mean of 52.1 [SD 8.82] versus control mean of 47.8 [SD 11.8]) and the mental health score (MCS) (intervention group mean of 47.3 [SD 8.65] versus control mean of 46.8 [SD 10.6]) without mentioning the significance level;
Jantzen 2009 reported that children and adolescents in the treatment group showed a significant increase in the social exclusion subscale in DISABKIDS, indicating better quality of life, based on postintervention scores (P value was not provided; d = 0.3 [Cohen] Cohen 1988);
May 2002 reported non‐significant postintervention mean between groups using the Short‐Form 36 mental component (intervention mean 43.69 [SD 11.51] versus wait‐list control mean 42.46 [SD 11.75]), and the physical component (intervention mean 50.39 [SD 9.37] versus wait‐list control mean 52.00 [SD 8.7]; P = 0.075);
Ring 2018 employed the Epilepsy and Learning Disabilities Quality of Life scale (ELDQoL). The comparison between the intervention group postintervention mean and control group was non‐significant (mood scale intervention mean 26.01 [SD 8.74] versus control mean 26.64 [SD 8.81]; behavior scale intervention mean 15.65 [SD 6.51] versus control mean 16.28 [SD 6.77]);
Rau 2006 used the self‐reported German questionnaire, Gesundheitsbezogene Lebensqualität und psychosoziale Auswirkungen der Epilepsie, (HRQoL and psychosocial consequences of epilepsy (intervention mean 70.62 [SD 13.29] versus wait‐list control mean 77.25 [SD 15.0]; P = 0.075).

Psychiatric comorbidities outcome measures

Depression

Sixteen of the 36 studies included the level of depression as an outcome measure; all of them indicated that there were no statistical differences between the treatment and control groups at baseline. We used postintervention means to compare the differences between the two groups. Eight studies reported a significant postintervention difference between the intervention and control groups (Ciechanowski 2010; Fraser 2015; Gandy 2014; Hum 2019; Martinović 2006; Schröder 2014; Tang 2015; Thompson 2010). Nine studies used more than one outcome measure (Caller 2016; Ciechanowski 2010; Gandy 2014; Gilliam 2019; Hum 2019; Martinović 2006; Meyer 2019; Orjuela‐Rojas 2015; Sajatovic 2016). Two studies, which used two outcome measures reported both significant and non‐significant results (Ciechanowski 2010; Hum 2019). Eight studies reported non‐significant results (Caller 2016; Dorris 2017; Leenen 2018; May 2002; Orjuela‐Rojas 2015; Pfäfflin 2016; Ridsdale 2018; Sajatovic 2016). A lower mean score indicated fewer depressive complaints.

Beck Depression Inventory and Beck Depression Inventory‐II (BDI and BDI‐II)

In the six studies that used the BDI or BDI‐II, four reported significantly better postintervention mean depressive symptoms in the treatment groups: Martinović 2006 (intervention mean 5.4 [SD 2.97] versus control mean 7.8 [SD 2.66]; P < 0.05); Schröder 2014 (intervention mean 15.84 [SD 13.00] versus control mean 18.37 [SD 10.23]; P = 0.01); Tang 2015 (intervention mean 6.90 [95% CI 4.49 to 9.31] versus control mean 9.47 [95% CI 6.26 to 12.67]; P = 0.045); Thompson 2010 (intervention mean 5.5 versus control mean 10.6; P value < 0.01). Gilliam 2019 and Orjuela‐Rojas 2015 reported a non‐significant difference (Gilliam 2019: intervention mean 12.8 [SD 11.9] versus control mean 12.3 [SD 9.9]; P value was not reported; Orjuela‐Rojas 2015: intervention mean 17.2 versus control mean 14.6; P = 0.58; SD was not reported).

Hospital Anxiety and Depression Scale (HADS) for assessing depression (HADS‐D)

Five studies used the HADS to assess depressive symptoms (Gandy 2014; Leenen 2018; Orjuela‐Rojas 2015; Pfäfflin 2016; Ridsdale 2018). Gandy 2014 reported significantly better postintervention depressive symptoms in the treatment group (intervention mean 4.58 [SD 3.59] versus control mean 5.50 [SD 5.26]; P = 0.048), while in Leenen 2018 (intervention mean 5.7 [SD 2.7] versus control mean 5.5 [SD 2.6]), the P value was non‐significant, without a precise value; Orjuela‐Rojas 2015 (intervention mean 5.4 versus control mean 5.2; P = 0.93; SD was not reported); Pfäfflin 2016 (intervention mean 9.0% ≥ 11 and control mean 5.5% ≥ 11; P = 0.432); and Ridsdale 2018 (intervention mean 5.5 [SD 3.9] versus control mean 5.0 [SD 3.9], P value was non‐significant, without precise value) reported a non‐significant difference.

Patient Health Questionnnaire‐9 for accessing depression (PHQ‐9)

Five studies used the PHQ‐9 to assess depressive symptoms (Caller 2016; Fraser 2015; Meyer 2019; Sajatovic 2016; Thompson 2010). Caller 2016 and Sajatovic 2016 reported a non‐significant difference in postintervention changes (Caller 2016: intervention mean change −0.7 [SD 1] versus control mean change 1.2 [SD 1.2]; Sajatovic 2016: intervention group mean 9.70 [SD 5.55] versus control mean 11.76 [SD 5.72], P = 0.25). Fraser 2015 and Meyer 2019 reported significantly lower depression scores in the treatment group (Fraser 2015: intervention mean 6.3 [SD 5.5] versus control mean 8.6 [SD 6]; P = 0.02; Meyer 2019: intervention mean 10.42 [SD 4.38] versus control mean 12.73 [SD 4.19]; P < 0.001). Thompson 2010 used the PHQ‐9 to identify individuals with a major depressive disorder at baseline, but did not provide postintervention data.

Neurological Disorders Depression Inventory for Epilepsy (NDDIE)

Four studies used the NDDIE to assess depressive symptoms (Caller 2016; Gandy 2014; Hum 2019; Meyer 2019). Gandy 2014 and Meyer 2019 found significantly reduced postintervention depressive symptoms in the treatment group (Gandy 2014: intervention mean 14.3 [SD 3.4] versus wait‐list control mean 16.48 [SD 3.81]; P = 0.045; Meyer 2019; intervention mean 14.35 [SD 3.4] versus control mean 16.02 [SD 3.15]; P > 0.001). Hum 2019 reported a significant decrease in scores across time for the intervention group (mean change from baseline: 1.75 [SD 0.8], P = 0.023) and the active control group (mean change from baseline: 1.71 [SD 0.8], P = 0.016), but not the wait‐list control group (mean change from baseline: 0.45 [SD 0.6], P = 0.654). Caller 2016 reported a non‐significant difference in depressive symptoms between groups (treatment mean change from baseline −0.4 [SD 0.6] versus control mean change from baseline 0.7 [SD 0.8]; P = 0.30.

Other depression outcome measures

Three studies that used other measures found significant effects in the treatment group for postintervention depressive symptoms. Martinović 2006 used the Center for Epidemiological Study on Depression scale (intervention mean 9.8 [SD 4.2] versus control mean 13.6 [SD 4.64]; P < 0.05) and the Hamilton Depression Scale (intervention mean 3.3 [SD 1.29] versus control mean 5.8 [SD 1.98]; P < 0.05). Hum 2019 used the Quick Inventory of Depressive Symptomatology (QIDS) and found a significant decrease in scores across time for the intervention group (P = 0.014) and the active control group (P = 0.02), but not for the wait‐list control group (P = 0.085). However, the significance level of the comparison of postintervention mean scores was not reported (intervention group: 9.55 [SD 1.1], active control group: 10.63 [SD 1.0] and wait‐list control group 10.73 [SD 1.5]). Sajatovic 2016 used the Montgomery and Asberg Depression Rating Scale (MADRS) and found a significant treatment‐by‐time interaction effect (P = 0.036), and the comparison of postintervention group means (intervention group mean 16.75 [SD 10.28] versus control group mean 22.94 [SD 11.81]) was non‐significant (P = 0.09).

Four studies that used other measures found non‐significant differences in depressive symptoms between groups. Ciechanowski 2010 used the Hopkins Symptom Checklist‐20 (treatment mean change from baseline −0.18 [SD 0.7] versus control mean change from baseline −0.48 [SD 0.7]; P = 0.09). May 2002 used the Depressive Mood Scale (intervention mean 13.63 [SD 8.99] versus wait‐list control mean 12.22 [SD 8.86]; the P value was non‐significant, without precise value). Brown 2019 and Dorris 2017 used pediatric depression outcome measures and found non‐significant differences in depressive symptoms between groups. Brown 2019 used the Children's Depression Inventory — Short (CDI‐S): intervention mean 45.6 (SD 6.9) versus control mean 44 (SD 4.8); Dorris 2017 used the Pediatric Index of Emotional Distress (PI‐ED): intervention group mean: 14.95 (SD 6.39) versus control group mean: 13.39 (SD 6.69), P value was non‐significant, without precise value.

Suicidal ideation

While Ciechanowski 2010 reported a significantly smaller proportion of participants with suicidal ideation at follow‐up (decreasing 24% in the intervention group and increasing 12% in the usual care group; P = 0.025; post‐treatment outcomes were not reported), Orjuela‐Rojas 2015 did not find a significant difference in suicide risk between groups, using the Mini International Neuropsychiatric Interview (intervention mean 1.1 versus control mean 0.6; P = 0.42; SD was not reported).

Anxiety

Eight studies included the level of anxiety symptoms as an outcome measure (Fraser 2015; Gandy 2014; Leenen 2018; Meyer 2019; Orjuela‐Rojas 2015; Pfäfflin 2016; Ridsdale 2018; Tang 2015). One study reported significant baseline differences between the intervention (11.2) and control (8.3); P = 0.04 (Orjuela‐Rojas 2015). We used postintervention means to compare the difference between the two groups. A lower mean score indicated fewer anxiety complaints. Two studies reported a significant postintervention difference between the intervention and control groups (Meyer 2019; Tang 2015).

Beck Anxiety Inventory (BAI)

Among the eight studies that examined anxiety symptoms, only one study, using the BAI, reported significantly fewer postintervention anxiety symptoms between groups (intervention mean 9.73 [95% CI 6.35 to 13.22) versus control mean 10.70 [95% CI 7.24 to 14.16]; P = 0.008 [Tang 2015]). None of the remaining studies, assessing anxiety with validated outcome measures, reported a significant difference.

Hospital Anxiety and Depression Scale (HADS) for assessing anxiety (HADS‐A)

Gandy 2014 reported a non‐significant postintervention difference between groups (intervention mean 6.11 [SD 2.96] versus control mean 7.45 [SD 3.78]; P = 0.089). Similar findings were also reported by Leenen 2018 (intervention mean 5.2 [SD 3.5] versus control mean 6.1 [SD 4.2], P value was non‐significant, without precise value); Pfäfflin 2016 (20.9% with HADS‐A ≥ 11 in the intervention group and 17.8% with HADS‐A ≥ 11 in the control group; P value was non‐significant, without precise value); Ridsdale 2018 (intervention mean 9.0 [SD 5.0] versus control mean 7.8 [SD 4.8], P = 0.917) and Orjuela‐Rojas 2015 (intervention mean 9.7 versus control mean 9.2; P = 0.8). It is worth noting that in this study, the treatment group had a significantly higher anxiety score at baseline compared to control (11.2 versus 8.3; P value = 0.04).

Generalized Anxiety Disorder‐7 (GAD‐7)

While Meyer 2019 found a significant difference between groups at postintervention in symptoms of anxiety using the Generalized Anxiety Disorder‐7 (intervention mean 7.74 [SD 4.3] versus control mean 9.82 [SD 3.91]; P < 0.001), Fraser 2015 did not find a significant difference between groups (intervention mean 5.4 [SD 6.6] versus control mean 6.1 [SD 5.1]; P = 0.282).

Seizure‐related outcomes

Fifteen studies included seizure‐related variables as outcome measures (Au 2003; Ciechanowski 2010; Gilliam 2019; Jantzen 2009; Leenen 2018; Lundgren 2006; Lundgren 2008; May 2002; Pakpour 2015; Rau 2006; Ridsdale 2018; Ring 2018; Sajatovic 2016; Sajatovic 2018; Tang 2015). Sajatovic 2018 reported negative health events (NHEs) which included seizure count but also other variables such as epilepsy‐related emergency room visits and injuries. All of them reported no evidence of baseline imbalance between groups except in Lundgren 2008, in which no statistics were provided, and baseline imbalance was indicated from the raw data (treatment group N = 10, seizure frequency = 414; active control group N = 8, seizure frequency = 33). However, this study reported that 50% of participants in both the intervention and the active control groups were seizure‐free at postintervention. Five studies reported a significant postintervention difference between the intervention and control groups (Lundgren 2006; Lundgren 2008; May 2002; Pakpour 2015; Tang 2015).

Three studies reported a significant postintervention reduction in seizure frequency between groups:
May 2002 (seizures per month in six months: intervention mean 2.77 [SD 1.64] versus control mean 2.74 [SD 1.62]; P < 0.041);
Lundgren 2006 (seizures in one month: intervention mean 0.71 [SD 0.91] versus control mean 6.00 (SD 3.91); P < 0.001);
Tang 2015 (seizures in six weeks: intervention mean 5.9 [95% CI 2.88 to 8.92] versus control mean 7.33 [95% CI 3.46 to 11.21]; P = 0.018).

Twelve studies reported non‐significant postintervention differences between groups in seizure frequency: (Au 2003; Ciechanowski 2010; Gilliam 2019; Jantzen 2009; Lundgren 2008; Rau 2006; Ridsdale 2018; Sajatovic 2016), seizure recency (Ridsdale 2018), seizure severity (Gilliam 2019: seizure calendar; Leenen 2018: National Hospital Seizure Severity Scale (NHS3): intervention group mean 7.3 [SD 8.1] versus control group mean 8.4 [SD 9.3], P value was non‐significant, without precise value; Ring 2018: Epilepsy and Learning Disabilities Quality of Life seizure severity scale (ELDQoL‐SSS): intervention group mean 22.48 [SD 9.55] versus control group mean 23.07 [SD 9.70], P = 0.875).

Pakpour 2015 reported a significant postintervention reduction in seizure severity using the Liverpool Seizure Severity Scale (intervention mean 47.24 [SD 17.41] versus control mean 58.09 [SD 21.75]; P < 0.05).
Tang 2015 found no significant postintervention changes using the Seizure Severity Index (intervention mean 2.55 [95% CI 2.06 to 3.03] versus control mean 2.91 [95% CI 2.44 to 3.38]; P > 0.05).
Sajatovic 2018 also found no significant postintervention change of seizure count (intervention mean −1.4 [SD 5.12] versus control mean 5.5 [SD 0.62]; P = 0.60) and non‐significant differences in seizure severity (intervention mean 19.52 [SD 8.3] versus control mean 17.91 [SD 7.8]; P = 0.06).

Lundgren 2006 and Lundgren 2008 measured seizure index (seizure frequency x seizure duration in seconds), which was not a validated outcome measure. With a comparable baseline seizure index, Lundgren 2006 reported significant postintervention reduction. Lundgren 2008 reported a significant difference in change scores between the two groups.

Discussion

Despite our broad operational definition, psychological treatments for people with epilepsy have been investigated in a relatively small number of randomized controlled trials. We found 36 RCTs that fit our operational definition, and investigated HRQOL as a primary or secondary outcome parameter.

Summary of main results

Primary outcome measure

The Quality of Life in Epilepsy Inventory (QOLIE‐10, QOLIE‐31, QOLIE‐31‐P, QOLIE‐48, QOLIE‐89) was the most commonly used outcome measure for health‐related quality of life (HRQOL), and was used in 25 studies. Results from the meta‐analysis (11 studies: Au 2003; Caller 2016; Ciechanowski 2010; Fraser 2015; Gandy 2014; Gilliam 2019; Helde 2005; Leenen 2018; Martinović 2006; Orjuela‐Rojas 2015; Tang 2015) found significant postintervention improvement for the total score and for six out of seven QOLIE‐31 subscales (emotional well‐being, energy and fatigue, overall QOL, seizure worry, cognitive functioning, and medication effects). The mean improvement in the total score and one subscale measuring overall QOL exceeded the minimally important change (MIC) threshold established by Borghs 2012 for a small effect size (d = 0.3 [Cohen]), indicating a clinically meaningful postintervention improvement in HRQOL (MIC: total score: 4.73 points; QoL score: 5.22 points). In the remaining 14 studies that were not included in the meta‐analysis, seven reported significant differences between the treatment and control groups at post‐treatment measures (Hosseini 2016; Lua 2013; Meyer 2019; Pakpour 2015; Sajatovic 2018; Turan Gurhopur 2018; Yadegary 2015), and seven reported non‐significant group differences at post‐treatment measures (Beretta 2014; DiIorio 2011; Pfäfflin 2016; Pramuka 2007; Ridsdale 2018; Sajatovic 2016; Schröder 2014). Eleven studies used HRQOL outcome measures other than the QOLIE inventories, but only three studies found a significant difference between the treatment and control groups at post‐intervention, indicating significantly better HRQOL in the treatment groups (Hum 2019; Jantzen 2009; Lundgren 2006).

Secondary outcome measures

Fewer than half of the studies (eight out of nineteen) examining depressive symptoms reported a decrease in symptoms for those receiving the treatment compared to control. Only two studies out of eight examining anxiety levels found a significant postintervention reduction in the treatment groups. A relatively small proportion of the studies (5/14) investigating seizure‐related outcome measures reported a significant reduction at postintervention in the treatment groups.

Overall completeness and applicability of evidence

The studies in this updated review evaluated complex psychological treatments, typically applied in tertiary‐care settings, and involved participant groups with comparable underlying epilepsy diagnoses, but differing severities of psychiatric and somatic comorbidities, and diverse cultural, ethnic, and socioeconomic backgrounds. There were differences between the included studies in their stated treatment methods, goals, strategies, and theoretical underpinnings. Psychologists with varying levels of experience delivered most of the treatments; team efforts including a wider range of specialists (doctors, nurses, social workers, etc.) were involved in some educational interventions. In some cases, the work of these therapists was carefully structured and supervised, while some studies presumably relied on shorter training courses.

We used 10 different outcome measures to investigate our primary outcome measure, i.e. HRQOL (QOLIE inventories, WHOQOL‐BREF, SWLS, DISABKIDS, SF‐12/‐36, PedsQL, GEOS‐YP, CheQoL, KIDSCREEN‐27, ELDQoL). Hence, our efforts to pool data were hampered to some extent by the wide diversity of outcome measures used in these trials. We addressed this by including as many comparable outcome as possible. We therefore sought raw data from authors that used the QOLIE inventories convertible to QOLIE‐31 (most commonly used in our included studies) and analyzed the mean change from baseline. In addition, we focused our meta‐analyses on skills‐based psychological interventions and did not include education‐only interventions. The studies which are included in meta‐analysis are representative of the whole evidence base for skills‐based psychological interventions. We have no reason to think that the results from these meta‐analyses would not be applicable in similar settings and patient groups.

Certainty of the evidence

We pooled results only from studies measuring the same construct (QOLIE‐31, QOLIE‐31‐P, QOLIE‐89) and limited meta‐analysis to fairly similar interventions, to avoid clinical heterogeneity. Overall, the certainty of evidence of the meta‐analysis was limited by a serious risk of bias in some of the included studies (e.g. Orjuela‐Rojas 2015, with four high‐risk and one unclear rating out of seven 'Risk of bias' parameters). Since most of the included studies had at least some bias issues, we did not perform a sensitivity analysis comparing studies with low risk of bias versus studies with high risk of bias. Given that the evidence directly answered our healthcare question, the results were precise enough, and fairly consistent across studies, and we found no evidence of publication bias or a dose‐response gradient in the included studies, we considered there was no reason to further downgrade the certainty of the evidence. As the effect was not large, neither was there reason to upgrade the certainty of the evidence. Interpreting these findings with the GRADE approach, we are therefore moderately confident in the effect estimate that skills‐based psychological interventions may enhance overall quality of life in people with epilepsy.

We identified RCTs and quasi‐RCTs to include in this review, based on our operational definition of psychological interventions, regardless of the nature of the control group. The heterogeneity of control groups (e.g. usual care, wait‐list control, other psychological or educational intervention, pharmacotherapy, etc.) may have influenced the interpretation of the effect of skills‐based psychological interventions for people with epilepsy because they may have been attenuated by another evidence‐based treatment (e.g. using sertraline for the treatment of depression in epilepsy in the control group, used by Gilliam 2019). As the number of included studies increases, we may take this into consideration and regroup our analyses in future updates.

In the process of evaluating the risks of bias, we considered two biases that are naturally inherent in the context of psychological intervention trials. The first bias was blinding of participants and the professionals who deliver the intervention. Although two studies managed to blind both participants and the professionals providing treatment (Martinović 2006 by using an active skills‐based intervention control group, and Ring 2018 by using a cluster‐randomized design), we considered this as a reasonable bias in psychological trials. Secondly, we rated all studies that included self‐selection, e.g. through advertisement, web‐based fora, study flyers, in the process of participant recruitment as having a high risk of bias for selective recruitment. We agreed that this is a high risk of bias for research design, but we could understand the reason from a clinical perspective, because patient volunteering may often reflect motivation to treatment. We therefore also considered this to be a reasonable bias in psychological trials.

Potential biases in the review process

The identification of relevant studies fitting our broad operational definition of psychological treatments was challenging. As outlined in the review protocol, we searched a wide variety of databases, including trial registers, and scanned reference lists of relevant systematic reviews. Two review authors independently evaluated all studies and referred to the wider group of review authors or the Epilepsy Review Group with any unresolved questions. Although this whole process was carried out carefully, we cannot discount the possibility that we may have missed a relevant study, or misjudged an included study or a study outcome. Since this review will be periodically updated, we will in future updates include any missed relevant studies, and correct misjudgments of included studies or study outcomes that come to our attention. Even though many of the included studies were published despite finding non‐significant results, we cannot discount the potential risk of publication bias. However, the research community did not make us aware of trials that had been stopped or not published because of non‐significant findings.

Three members of our review author team had also co‐authored three of the included studies (Gandy 2014; Ridsdale 2018; Tang 2015). Following the strict standards of the Cochrane review process, we felt that this contributed a necessary critical expertise with the implementation of RCTs in this area of research, rather than an increased risk of bias. Gandy and Goldstein were not involved in the actual review process. Since Tang had been involved in the review process, a second author with no conflict of interest (MR) assessed eligibility and risks of bias for her study.

Agreements and disagreements with other studies or reviews

The results of this review reinforced the conclusions of a recent systematic review of psychological treatments for epilepsy, which suggested that cognitive behavioral therapy and mindfulness‐based interventions had consistently demonstrated significant improvements in HRQOL in prospective uncontrolled, as well as in controlled study designs (Tang 2014).

This review is in keeping with a systematic review of cognitive behavioral therapy for depression in people with epilepsy, which suggested that interventions tailored toward improving depression were possibly efficacious (Gandy 2013). Our results are also in line with the previous Cochrane Review focusing on seizure frequency as a primary outcome parameter, in that we could draw no reliable conclusions about the efficacy of psychological treatments in controlling seizures (Ramaratnam 2008).

From our review of randomized controlled trials investigating psychological treatments, we have identified the following two research problems in this area.

The feasibility of using randomized trial designs to study psychological interventions has been repeatedly challenged, primarily because the possibility of realistically balancing prognostic factors, on average, across intervention groups is questionable. The psychological (and psychopathological) make‐up of most participants is regarded as being too multifaceted for this endeavor to be successful. In addition, these trial designs usually include pre‐ and postintervention outcome measurements only and are therefore limited in their ability to capture process‐related outcomes. (Tschuschke 2005). Technology‐focused outcome measures, such as electronic monitors and ecological momentary assessment (EMA) may provide systematic (i.e. equidistant) and frequent (e.g. daily) response data on symptoms, mood, and behavior. Such process‐oriented variables may provide insight in nonlinear and complex dynamics of human change processes and therefore allow for more individualized psychotherapeutic management (Modi 2017).

Another important contributing factor to the effectiveness of psychological treatments is the therapeutic relationship between recipient and therapist (Baldwin 2013). However, no RCT investigating psychological treatment for epilepsy included this relationship as a variable.

Figure 1

Study flow diagram.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 2

'Risk of bias' graph: review authors' judgements about each 'risk of bias' domain presented as percentages across all included studies

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 3

'Risk of bias' summary: review authors' judgements about each 'risk of bias' domain for each included study

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 4

Forest plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.1 QOLIE‐31‐ total score.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 5

Forest plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.2 QOLIE‐31 ‐ overall QoL subscale.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 6

Funnel plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.1 QOLIE‐31‐ total score.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Figure 7

Funnel plot of comparison: 1 QOLIE‐31‐ Comparison of mean change from baseline, outcome: 1.2 QOLIE‐31 ‐ overall QoL subscale.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.1

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 1: QOLIE‐31‐ total score

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.2

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 2: QOLIE‐31 ‐ overall QoL subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.3

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 3: QOLIE‐31 ‐ energy and fatigue subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.4

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 4: QOLIE‐31 ‐ emotional well‐being subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.5

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 5: QOLIE‐31 ‐ seizure worry subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.6

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 6: QOLIE‐31 ‐ cognitive functioning subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.7

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 7: QOLIE‐31 ‐ medication effects subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Analysis 1.8

Comparison 1: QOLIE‐31‐ Comparison of mean change from baseline, Outcome 8: QOLIE‐31 ‐ social function subscale

Ir a la figura de la revisiónAbrir en una pestaña nueva

Summary of findings 1. Psychological treatments compared with usual or supportive care

Psychological treatments compared with usual or supportive care
Patient or population: adolescents and adults with epilepsy Setting: outpatient clinic or outpatient clinic and by phone or in‐home sessions and by phone Intervention: skills‐based psychological interventions Comparison: wait‐list control (WLC), usual care (UC) or supportive care (SC) or antidepressant drug treatment
Outcomes	*Comparative effect sizes^ (95% CI)**		№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
Outcomes	Wait‐list control, usual care, supportive care or antidepressant drug treatment	Psychological treatments	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
QOLIE‐31 total score^a	The range of mean change in the control groups was −1.9 to 15.96 points.	The range of mean change in the intervention groups was 3.27 to 17.2 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.23 higher (95% CI 3.02 to 7.44 higher) than the control groups	643 (11 RCTs)	⊕⊕⊕⊝ MODERATE^c	2 out of 3 studies that could not be included in meta‐analysis due to use of QOLIE‐89 or QOLIE‐31‐P reported significantly higher postintervention QOLIE total scores in the treatment over the control groups (Hosseini 2016; Yadegary 2015). For narrative synthesis of all other HRQOL outcomes see Table 2.
QOLIE‐31 emotional well‐being subscale^a	The range of mean change in the control groups was −6.23 to 24.95 points.	The range of mean change in the intervention groups was 0.91 to 20.57 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.96 higher (95% CI 0.70 to 9.21 higher) than the control groups	643 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 energy or fatigue subscale^a	The range of mean change in the control groups was −5.3 to 17.69 points.	The range of mean change in the intervention groups was 0.44 to 18.75 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.25 higher (95% CI 1.56 to 8.93 higher) than the control groups	642 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 overall QoL subscale^a	The range of mean change in the control groups was −2.63 to 15 points.	The range of mean change in the intervention groups was 0.13 to 19.64 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 5.95 higher (95% CI 3.05 to 8.85 higher) than the control groups	639 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 seizure worry subscale^a	The range of mean change in the control groups was −5.18 to 17.26 points.	The range of mean change in the intervention groups was 2.74 to 28.56 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.35 higher (95% CI 1.35 to 7.35 higher) than the control groups	632 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 cognitive functioning subscale^a	The range of mean change in the control groups was −2.71 to 13.17 points.	The range of mean change in the intervention groups was 2.28 to 16.16 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 4.18 higher (95% CI 1.82 to 6.54 higher) than the control groups	641 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 medication effects subscale^a	The range of mean change in the control groups was −8.11 to 12.04 points.	The range of mean change in the intervention groups was 0.93 to 6.64 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 3.16 higher (95% CI 0.01 to 6.32 higher) than the control groups	643 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
QOLIE‐31 social function subscale^a	The range of mean change in the control groups was −4.28 to 13.98 points.	The range of mean change in the intervention groups was 2.3 to 10.49 points. The pooled mean change from baseline in the intervention groups measured at postintervention^b was on average 3.09 higher (95% CI ‐0.17 lower to 6.35 higher) than the control groups	630 (10 RCTs)	⊕⊕⊕⊝ MODERATE^c	‐
* Comparative effect sizes were calculated from the mean changes between baseline and post‐intervention in the intervention and control groups. CI: Confidence interval; QOLIE: Quality of life in epilepsy; RCT: randomized controlled trial
GRADE Working Group grades of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect. Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect.
^aRange 0 ‐ 100 points, higher score means higher quality of life. ^bThe median postintervention measurement point was 3 months (8 weeks to 2 years). ^cSerious risk of bias, i.e. included studies share serious risk of performance bias and five included studies share serious risk of attrition bias.

Summary of findings 1. Psychological treatments compared with usual or supportive care

Ir a la tabla de la revisión

Table 1. Intervention methods, strategies, and treatment goals

	Study (intervention acronym)	Main treatment method	Primary treatment goal	Main treatment strategy	Provider	Setting	Delivery	Timing	Participants
Skills‐based psychological interventions	Au 2003	Cognitive behavioral therapy	Seizure frequency	Stress management, cognitive restructuring, communication skills	Clinical psychologist	Clinic	Group	8 weekly 2‐hour sessions	N = 17 adults with at least 2 seizures per month, with subjectively reported psychological distress
	Ciechanowski 2010 (PEARLS)		Depressive symptoms	Cognitive restructuring to address negative depressive thinking + behavioral activation	Trained social worker	Home‐based + telephone calls	Individual	8 50‐min in‐home sessions in 5 months + 7 monthly 5‐ to 10‐min telephone calls	N = 80 adults with epilepsy with significant depression
	Gandy 2014				Intern psychologist	Clinic	Individual	1 x 1‐ to 2‐hour assessment session + 8 weekly 1‐hour sessions	N = 59 adults with epilepsy
	Gilliam 2019			CBT based on standardized and manual‐based Beck guidelines	Nurse educator and trained lay person with epilepsy	Therapist office	Individual	1‐hour session per week for 16 weeks	N = 98 adults (age 21 ‐ 75) with epilepsy and current major depressive episode
	Hum 2019 (UPLIFT)			see Thompson 2010	Licensed mental health professional and trained layperson with epilepsy	Telephone calls	Group	8 weekly 1‐hour sessions	N = 55 adults with epilepsy and depressive symptoms
	Martinović 2006			Cognitive restructuring to address negative depressive thinking + behavioral activation	NR	Clinic	Group	8 weekly sessions + 4 monthly sessions	N = 32 adolescents with epilepsy and subthreshold depression
	Meyer 2019 (Emyna)			Cognitive restructuring to address negative depressive thinking + behavioral activation	NA	Internet‐based	Individual	5 modules with no fixed sequence, each lasting for 60 ‐ 180 min	N 154 adult (> 18) with active epilepsy and a current diagnosis of moderate depression
	Orjuela‐Rojas 2015				Licensed CBT therapist and psychiatrist	Clinic	Group	12 weekly 90‐min sessions	N = 15 adults with epilepsy and major depression
	Schröder 2014 (Deprexis)				NA	Internet‐based	Individual	9 weekly modules (10 ‐ 60 min)	N = 78 adults with self‐reported depressive symptoms
	Thompson 2010 (UPLIFT)				Master of Public Health student and trained lay person with epilepsy	Internet‐based + telephone calls	Group	8 weekly 1‐hour sessions	N = 53 adults with epilepsy and depression (but not severe depression)
	Dorris 2017	Self‐management program	Quality of life	Medical self‐management and sleep hygiene, coping strategies and problem‐solving techniques based on CBT and mindfulness	Epilepsy nurse and clinical psychologist	Clinic	Group	6 weekly 120‐min sessions	N = 69 children and adolescents aged 12 ‐ 17 with epilepsy
	Fraser 2015 (PACES)		Self‐management	Medical and psychosocial self‐management + epilepsy‐related communication	Psychologist and trained lay person with epilepsy	Clinic	Group	8 weekly 75‐min sessions	N = 83 adults with epilepsy
	Leenen 2018 (ZMILE)		Self‐management and quality of life	Self‐monitoring, risk‐evaluation and management; shared decision‐making, goal‐setting skills	Nurse practitioner	Clinic	Group	5 weekly 2‐hour sessions followed by a 2‐hour booster session after 3 weeks	N = 87 adults with epilepsy and on AEDs
	Sajatovic 2016 (TIME)		Depressive symptoms	Personal goal‐setting exercises (with focus on coping with mental illness and epilepsy), stress management, and training to communicate with care providers	Nurse educator and trained lay person with epilepsy	Clinic	Group	12 weekly 60‐ to 90‐min sessions	N = 35 adults with epilepsy and comorbid mental illness
	Sajatovic 2018		Negative health events	SMART "self‐management for people with epilepsy and a history of negative health events"	Nurse educator and trained lay person with epilepsy	Clinic + telephone intervention calls + telephone maintenance	Group + individual	1 face‐to‐face 60‐ to 90‐min group; 7 Internet‐based group; 6 10‐ to 15‐min telephone maintenance	N = 111 adults with at least 1 negative health event within the past 6 months
	Yadegary 2015		Quality of life	Medical and psychosocial self‐management + seizure communication	NR	Clinic	Group	4 weekly 120‐min sessions	N = 60 adults with epilepsy
	DiIorio 2011 (WebEase)	Motivational interviewing (MI)	Medication adherence + perceived stress	Medication adherence + stress and sleep management	NA	Internet‐based	Individual	3 bi‐weekly modules	N = 194 adults with epilepsy
	Hosseini 2016		Quality of life	Enhancement of internal motivation for change, by overcoming dualism	Psychologist and trained layperson with epilepsy	Clinic	Group	5 sessions in 20 days	N = 56 adults with epilepsy.
	Pakpour 2015		Medication adherence	MI techniques	Health psychologist	Clinic	Individual	3 weekly 40‐ to 60‐min sessions	N = 275 adults with epilepsy
	Lundgren 2006; Lundgren 2008	Mindfulness therapy (MT)	Quality of life	ACT + seizure management	Clinical psychologist	Clinic	Group + individual	5 individual 90‐min sessions + 2 x group 3‐hour sessions + 2 x 1‐hour boosters at 6 and 12 months	N = 27 (Lundgren 2006) N = 18 adults with epilepsy (Lundgren 2008)
	Tang 2015	Mindfulness therapy (MT)	Quality of life	Epilepsy management + mindfulness techniques + seizure‐related acceptance	Clinical psychologist	Clinic	Group	4 x bi‐weekly 2 x.5‐hour sessions	N = 61 adults with drug‐resistant epilepsy
	Brown 2019	Behavior‐change counseling	Physical activity and quality of life	Self‐regulatory skills to support behavior change	Trained research assistant	Clinic	Individual	15‐min sessions: weekly/bi‐weekly/monthly weeks 1 – 4/ 6 – 12/16 – 24	N = Children aged 8 – 14 years with epilepsy
	Caller 2016 (HOBSCOTCH)	Cognitive, memory + self‐management training	Quality of life	Problem‐solving therapy and behavior modification strategies + seizure management + social skills	Specialized nurse	Home‐based + telephone calls	Group + individual	8 weekly 40‐ to 60‐min sessions	N = 66 adolescents and adults with epilepsy and self‐reported memory complaints
	Helde 2005	Epilepsy education + nurse‐led counseling	Quality of life	Personalized counseling + disease knowledge + drug adherence	Specialized nurse	Clinic + phone calls	Group + individual	1‐day group + phone calls every 3 months for 2 yrs	N = 114 adults with epilepsy
	Pramuka 2007	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Psychologist and epilepsy nurse	Clinic	Group	6 weekly 2‐hour sessions	N = 55 adults with epilepsy
	Ring 2018	Learning Disability Epilepsy Specialist Nurse Competency Framework	Seizure frequency and quality of life	Provide care according to guidelines developed by the UK ESNA and UK Royal College of Nursing	Licensed mental health professional and trained lay person with epilepsy	Home visits, telephone, clinics and visits to the local primary care or ID team base	Individual	On an as‐needed basis for 24 weeks	N = 312 adults with epilepsy and intellectual disability
Education‐only interventions	Beretta 2014 (EDU‐COM)	Patient‐tailored medication education	Drug‐related problems	Personalized education on drug interaction and tolerability	Treating physician	Clinic	Individual	1‐hour session + booster session after 1 month	N = 174 adults with epilepsy and chronic comorbidity
	Edward 2019	Epilepsy education program	Seizure frequency	Education program developed based on the self‐determination theory (managing epilepsy and medical care; socializing on a budget, healthy lifestyle, emotional management)	Specialized nurse	Not specified in the publication	Not specified in the publication	1 x 120‐min session	N = 35 adults with epilepsy
	Jantzen 2009 (FLIP&FLAP )	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Trained nurses, social workers, medical doctors or psychologists	Clinic	Group	2‐day course (14 hours)	N = 192 children and adolescents with epilepsy, including parents
	Lua 2013	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	NR	SMS‐based	Individual	11 weekly modules	N = 144 adults with epilepsy
	May 2002 (MOSES)	Epilepsy education program	Quality of life	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Trained nurses, social workers, medical doctors or psychologists	Clinic	Group	2‐day course (14 hours)	N = 383 adolescents and adults with epilepsy
	Pfäfflin 2016	Counseling	Satisfaction with information and support	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	Specialized nurse	Clinic	Individual	Delivery during routine visits	N = 187 adults with epilepsy
	Rau 2006 (FAMOSES)	Epilepsy education program	Knowledge + coping	Disease knowledge, advocacy topics, self‐management, psychosocial aspects	NR	Clinic	Group	2‐day course (14 h)	N = 70 children with epilepsy
	Ridsdale 2018 [SMILE (UK)]	Epilepsy education program (May 2002)	Quality of life	see May 2002	Nurse educator and trained lay person with epilepsy	Clinic	Group	2‐day course (16 h)	N = 314 adolescents (≥ 16 years) and adults with poorly‐controlled epilepsy
	Turan Gurhopur 2018	Epilepsy education program	Epilepsy‐specific knowledge, self‐efficacy, quality of life	Modular education program including epilepsy knowledge, seizure management, and social aspects of epilepsy	NR	Clinic	Individual	2 ‐ 3 days with a total of 16 hours	N = 92 including children with epilepsy aged 7 ‐ 18; and parents of children with epilepsy
ACT: acceptance and commitment therapy; AED: anti‐epilepsy drug; CBT: cognitive behavioral therapy; ESNA: EpilepSy Nurses Association; ID: intellectual disability; MI: motivational interviewing

Table 1. Intervention methods, strategies, and treatment goals

Ir a la tabla de la revisión

Table 2. Effects of interventions

	Study (intervention acronym)	HRQOL	Depression	Anxiety	Seizure‐related outcomes	Additional outcomes	Time points measured
Skills‐based psychological interventions	Au 2003	QOLIE‐31^b	NA	NA	seizure frequency^a,d	ESES	1) baseline 2) postintervention
	Brown 2019	CHEQOL^a,d, KIDSCREEN‐27^a,d	CDI‐S^a,d	NA	NA	Physical activity^a	1) baseline 2) postintervention (28 after baseline) 3) 52‐week follow‐up
	Caller 2016 (HOBSCOTCH)	QOLIE‐31^a,b	PHQ‐9^d NDDI‐E^d	NA	NA	Self‐reported cognitive and executive function	1) baseline 2) postintervention
	Ciechanowski 2010 (PEARLS)	QOLIE‐31^b	HSCL‐20^a,d suicidal ideation^a,c	NA	seizure frequency^d	NA	1) baseline 2) postintervention 3) 12‐month follow‐up 4) 18‐month follow‐up
	DiIorio 2011 (WebEase)	QOLIE‐10^d	NA	NA	NA	ESI‐R^a, ESMS^a, MAS^a, PSQI^a, PSS^a, Epilepsy Knowledge Profile	1) baseline 2) postintervention 3) 12‐week follow‐up
	Dorris 2017	PedsQL^a,d GEOS‐YP^a,d	PI‐ED^d	NA	NA	EKP‐G, SSEC‐C, B‐IPQ	1) baseline 2) postintervention 3) 3‐month follow‐up 4) 6‐month follow‐up
	Fraser 2015 (PACES)	QOLIE‐31^b	PHQ‐9^c	GAD‐7^d	NA	ESES^a, ESMS^a	1) baseline 2) postintervention 3) 6‐months follow‐up
	Gandy 2014	QOLIE‐31^b	HADS‐D^a,c NDDI‐E^a,c	HADS‐A^a,d	NA	NA	1) baseline 2) postintervention 3) 3‐month follow‐up
	Gilliam 2019	QOLIE‐89^b	BDI II^{a, d} CES‐D^{a, d}	NA	Focal impaired seizures/month^d , Generalized tonic‐clonic seizures/month^d	Adverse events profile^d	1) baseline 2) 8‐week interim assessment 3) 16‐week (postintervention)
	Helde 2005	QOLIE‐89^a.b	NA	NA	NA	VAS scale	1) baseline 2) postintervention
	Hosseini 2016	QOLIE‐89^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
	Hum 2019 (UPLIFT)	WHOQOL‐BREF^a,d	QIDS^a,c NDDIE^a,d	NA	NA	NA	1) baseline, 2) 6‐month follow‐up 3) 12‐month follow‐up
	Leenen 2018 (ZMILE)	QOLIE‐31‐P^b,d	HADS‐D^d	HADS‐A^d	NHS3^d	GSES^a, GSES^a, MEMS, MARS, UPCC, SIDAED	1) baseline 2) 3‐month follow‐up 3) 6‐month follow‐up
	Lundgren 2006	SWLS^a WHOQOL‐BREF^a	NA	NA	seizure frequency^c seizure index^c	NA	1) baseline 2) postintervention 3) 6‐month follow‐up 4) 12‐month follow‐up
	Lundgren 2008	SWLS^a, WHOQOL‐BREF^a	NA	NA	seizure frequency^d, seizure index^c	NA	1) baseline 2) postintervention 3) 6‐month follow‐up 4) 12‐month follow‐up
	Martinović 2006	QOLIE‐31^b	BDI^a,d CES‐D^a,d HAMD^a,c	NA	NA	NA	1) baseline 2) postintervention 3) 9‐month follow‐up
	Meyer 2019	QOLIE‐10^c	PHQ‐9^{a, c}, NDDIE^c DASS^c WSAS^c	GAD‐7^c	NA	NA	1) baseline 2) 3‐month (postintervention) 3) 6‐month follow‐up 4) 9‐month follow‐up
	Orjuela‐Rojas 2015	QOLIE‐31^b	BDI^a,c HADS‐D^a,d MINI^a,d	HADS‐A^a,d	NA	NA	1) baseline 2) mid‐intervention 3) postintervention
	Pakpour 2015	QOLIE‐31^c	NA	NA	LSSS^d	MARS^a; for additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention 3) 6‐month follow‐up
	Pramuka 2007	QOLIE‐89^a,d	NA	NA	NA	ESES, WPSI, LOC	1) baseline 2) postintervention 3) 6‐month follow‐up
	Ring 2018	ELDQoL‐SSS^d	NA	NA	ELDQoL‐SSS^a,d	For additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Sajatovic 2016 (TIME)	QOLIE‐10^d	MADRS^a,d PHQ‐9^d	NA	seizure frequency^d	BPRS, GAF, WHODASII, PSQI, ESES, MSPSS, ISMI, ESS	1) baseline 2) postintervention 3) 16 week‐follow up
	Sajatovic 2018	QOLIE‐10^c SF‐36 PCS^c SF‐36 MCS^c	PHQ‐9^c MADRS^c	NA	LSSS^c	ESES^c, MSPSS^d, ESMS^c, ESS^{d, NHEa}	1) baseline 2) 10‐week follow‐up 3) 24‐week follow‐up
	Schröder 2014 (Deprexis)	QOLIE‐31^d WHOQOLBREF^d	BDI^a,c	NA	NA	NA	1) baseline 2) postintervention
	Tang 2015	QOLIE‐31‐P^a,b	BDI‐II^c	BAI^c	seizure frequency^c SSI^d	NA	1) baseline 2) postintervention
	Thompson 2010 (UPLIFT)	SWLS^d	BDI^a,c	NA	NA	DCSES, SCS, knowledge and skills, BRFSS	1) baseline 2) postintervention 3) 4‐month follow‐up
	Yadegary 2015	QOLIE‐31‐P^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
Education‐only interventions	Beretta 2014 (EDU‐COM)	QOLIE‐31^d	NA	NA	NA	Drug‐related problems^a	1) baseline 2) postintervention 3) 6‐month follow‐up
	Edward 2019	SF12‐PCS (ns) SF12‐MCS (ns) SWLS (ns)	NA	NA	NA	CD‐RISC (ns), MMAS‐8 (ns)	1) baseline 2) 6‐month follow‐up postintervention
	Jantzen 2009 (FLIP&FLAP )	DISABKIDS^c	NA	NA	Seizure‐free episode^d	Disclosure of epilepsy	1) baseline 2) postintervention
	Lua 2013	MQOLIE‐30^a,c	NA	NA	NA	NA	1) baseline 2) postintervention
	May 2002 (MOSES)	SF‐36^a,d	Depression Scale D‐S^a,d	NA	Seizure frequency^c	For additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Pfäfflin 2016	QOLIE‐31 (Overall QoL)^d	HADS‐D^d	HADS‐A^d	NA	Satisfaction with information and support^a	1) baseline 2) postintervention
	Rau 2006 (FAMOSES)	KINDL^d	NA	NA	Seizure frequency^d	Epilepsy knowledge^a; for additional outcomes please see Characteristics of included studies table	1) baseline 2) postintervention
	Ridsdale 2018 (SMILE [UK])	QOLIE‐31‐P^a,d	HADS‐A^d	HADS‐A^d	seizure frequency and seizure recency^d	Impact of epilepsy, stigma of epilepsy, medication adherence, self‐mastery and control	1) baseline 2) 6‐month follow‐up 3) 12‐month follow‐up
	Turan Gurhopur 2018	QOLIE‐48^c	NA	NA	NA	SSES‐C^c, EKTC^c	1) baseline (pretest), 2) immediately after intervention (post‐test 1), 3) 1‐month follow‐up 4) 3‐month follow‐up
^aprimary outcome measure(s) in study. ^bincluded in meta‐analysis. Interpretation of post‐intervention outcomes ^cSignificant improvement in treatment group when comparing post‐intervention outcomes of treatment and control group. ^dNo significant difference between treatment and control group at postintervention based on mean comparisons. NA: not applicable ns ‐ Not specified, with no information in the publication suggesting significant difference between treatment and control group at post‐intervention based on mean comparisons Outcome Measures BAI ‐ Beck Anxiety Inventory; BDI or BDI II ‐ Beck Depression Inventory or Beck Depression Inventory II; B‐IPQ ‐ Brief ‐ Illness Representations Questionnaire; BPRS ‐ Brief Psychiatric Rating Scale; BRFSS ‐ Behavioral Risk Factor Surveillance System; CES‐D ‐ Center for Epidemiological Study on Depression scale; CDI‐S ‐ Children's Depression Inventory ‐ Short; CD‐RISC ‐ Connor‐Davidson Resilience Scale; CHEQOL ‐ Childhood Epilepsy Quality of Life scale; DASS21 ‐ Depression Anxiety Stress Scale‐21; DISABKIDS ‐ Modular HRQOL questionnaire; DCSES ‐ Depression Coping Self‐Efficacy Scale; EKP‐G ‐ Epilepsy Knowledge Profile‐General; EKTC ‐ Epilepsy Knowledge Test for Children; ELDQoL‐SSS ‐ Epilepsy and Learning Disabilities Quality of Life Seizure Severity Scale; ESES ‐ Epilepsy Self‐Efficacy Scale; ESMS ‐ Epilepsy Self‐Management Scale; ESS ‐ 10‐item Epilepsy Stigma Scale; ESI‐R ‐ Revised Epilepsy Stressor Inventory; GAD‐7 ‐ Generalized Anxiety Disorder‐7; GAF ‐ Global Assessment of Functioning; GEOS‐YP ‐ Glasgow Epilepsy Outcome Scale for Young Persons; GSES ‐ General Self‐Efficacy Scale; HADS ‐ Hospital Anxiety Depression Scale; HAMD ‐ Hamilton Depression Scale; ISMI ‐ Internalized Stigma of Mental Illness Scale; HSCL‐20 ‐ Hopkins Symptom Checklist‐20; KINDL ‐ Gesundheitsbezogene Lebensqualität und psychosoziale Auswirkungen der Epilepsie (Health‐related Quality of Life and psychosocial consequences of epilepsy); LOC ‐ Locus of Control Scale; LSSS ‐ Liverpool Seizure Severity Scale; MADRS ‐ Montgomery‐Asbery Depression Rating Scale; MARS ‐ Medication Adherence Report Scale; MAS ‐ Medication Adherence Scale; MEMS ‐ Medication Event Monitoring System; mBDI ‐ Modified Beck Depression Inventory; MCMI‐III ‐ Millon Clinical Multiaxial Inventory‐III; MINI ‐ Mini International Neuropsychiatric Interview; MMAS‐8 ‐ Morisky Medication Adherence Scale (MMAS‐8); MQOLIE‐30 ‐ Malay Quality of Life Inventory in Epilepsy‐30; MSPSS ‐ Multidimensional Scale of Perceived Social Support; NDDI‐E ‐ Neurological Depressive Disorders Inventory‐Epilepsy; NHS3 ‐ National Hospital Seizure Severity Scale; PedsQL ‐ Paediatric Quality of Life Inventory PedsQL™; PHQ‐9 ‐ Patient Health Questionnaire‐9; PI‐ED ‐ Paediatric Index of Emotional Distress; PSQI ‐ Pittsburgh Sleep Quality Index; PSS ‐ Perceived Stress Scale; QOLIE‐31, QOLIE‐31‐P, QOLIE‐89 ‐ Quality of Life in Epilepsy‐31, Patient‐weighted Quality of Life in Epilepsy‐31, Quality of Life in Epilepsy‐89; SCS ‐ Self‐compassion Scale; SF12‐PCS ‐ Short‐Form 12 Physical Health Score; SF12‐MCS ‐ Short Form 12 Mental Health Score; SF‐36 ‐ Short‐Form 36; (PCS ‐ physical health score, MCS ‐ mental health score); SIDAED ‐ Side‐effects of Antiepileptic Drugs; SSEC‐C ‐ Seizure Self Efficacy Scale for Children; SSI ‐ Seizure Severity Index; SWLS ‐ Satisfaction with Life Scale; UPCC ‐ Utrecht Proactive Coping Competence; VAS scale (Helde 2005) ‐ General satisfaction with the follow‐up by the Neurological Clinic during the last 2 years; WHODASII ‐ World Health Organization Disability Assessment Schedule II; WHOQOL‐BREF ‐ World Health Organization Quality of Life instrument, short version; WPSI ‐ Washington Psychosocial Seizure Inventory; WSAS ‐ Work and Social Adjustment Scale; 4‐point Likert scale (Martinović 2006) ‐ Rating of positive and negative thoughts

Table 2. Effects of interventions

Ir a la tabla de la revisión

Comparison 1. QOLIE‐31‐ Comparison of mean change from baseline

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 QOLIE‐31‐ total score Show forest plot	11	643	Mean Difference (IV, Random, 95% CI)	5.23 [3.02, 7.44]

1.2 QOLIE‐31 ‐ overall QoL subscale Show forest plot	10	639	Mean Difference (IV, Random, 95% CI)	5.95 [3.05, 8.85]

1.3 QOLIE‐31 ‐ energy and fatigue subscale Show forest plot	10	642	Mean Difference (IV, Random, 95% CI)	5.25 [1.56, 8.93]

1.4 QOLIE‐31 ‐ emotional well‐being subscale Show forest plot	10	643	Mean Difference (IV, Random, 95% CI)	4.96 [0.70, 9.21]

1.5 QOLIE‐31 ‐ seizure worry subscale Show forest plot	10	632	Mean Difference (IV, Random, 95% CI)	4.35 [1.35, 7.35]

1.6 QOLIE‐31 ‐ cognitive functioning subscale Show forest plot	10	641	Mean Difference (IV, Random, 95% CI)	4.18 [1.82, 6.54]

1.7 QOLIE‐31 ‐ medication effects subscale Show forest plot	10	643	Mean Difference (IV, Random, 95% CI)	3.16 [0.01, 6.32]

1.8 QOLIE‐31 ‐ social function subscale Show forest plot	10	630	Mean Difference (IV, Random, 95% CI)	3.09 [‐0.17, 6.35]

Comparison 1. QOLIE‐31‐ Comparison of mean change from baseline

Ir a la tabla de la revisión

	Idioma de la Revisión Cochrane Escoja su idioma de preferencia para las revisiones Cochrane y otros contenidos. Las secciones sin una traducción aparecerán en inglés.

	Idioma de la web Escoja su idioma de preferencia para la web de la Biblioteca Cochrane.

Idioma de la Revisión Cochrane

Idioma de la web

Resumen

Antecedentes

Objetivos

Métodos de búsqueda

Criterios de selección

Obtención y análisis de los datos

Resultados principales

Conclusiones de los autores

PICO

PICO

Population

Intervention

Comparison

Outcome

Resumen en términos sencillos

Tratamientos psicológicos para los pacientes con epilepsia

Resumen visual

Authors' conclusions

Implications for practice

Implications for research

Increasing overall quality of reporting

Increasing comparability by using common and meaningful HRQOL outcome measures

Increasing overall quality of study designs

Increasing overall generalizability

Increasing implementation efforts

Summary of findings

Background

Description of the condition

Description of the intervention

1. Skills‐based psychological interventions

2. Education‐only interventions

How the intervention might work

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Main outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

References from published studies and relevant systematic reviews

Other sources

Data collection and analysis

Selection of studies

Data extraction and management

Study methods

Participants

Outcome data

Results

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Results

Description of studies

Results of the search

Included studies

Interventions

1. Skills‐based psychological interventions (27 studies, 2240 participants)

2. Education‐only interventions (9 studies, 1286 participants)

Intervention delivery

Control groups

Outcome measures

Health‐related quality of life

Psychiatric comorbidities: depression and anxiety

Seizure‐related outcomes